Low Cost Dust Collectors vs Shop-Vac

I was using a shop vacuum for my 6040 CNC router dust collection, but it was loud, and my almost new Craftsman XSP blew it’s guts out from running all night. I changed to a small Shop Vac brand one, but I knew it was only a matter of time because these motors are just not meant for long-running use. I decided to get something more purpose-built.

Harbor Freight had two Dust collectors. One was listed as “1 HP” and $169.99. The other was “2 HP” and $299.99 – but I was tempted to get the larger one. Still, I got the smaller one for two reasons. One was that the larger one said “20 amps” and I didn’t have that much spare power in this room (though in hindsight, it it might only be 10 amps). The other is that it was a massive floor-standing unit and this room was already fairly crowded. It thought the smaller one would be powerful enough because it said: “The 1 HP portable dust collector produces 10 times more suction than ordinary shop vacuums.” 10x is a lot, so that would be great.

When I got home and unboxed it, I thought it looked very nice. I turned it on, and it was for sure quieter than my shop vac. The suction though was not much at all. It subjectively felt about 1x the suction of a shop vac. So, to see if it was really a 1 HP motor, I used a watt meter. It drew only about 400 watts. There are 746 watts in a HP, and motors are 50-90 percent efficient, so assuming it was 80% efficient (which is generous), it was about 0.43 HP. I could forgive them if they called it 1/2 HP, but a motor that is just over 1/3 HP is too far from 1 HP to be sold as such. Between the false HP marketing and the “10x” the suction, I felt mislead and decided to return it.

I then searched Amazon for what else was available. I found two WEN units. Having just purchased their electric planer and being happy with the quality and the very low price, I knew I had to try one. But which one? One of them was listed as “WEN DC3401 5.7-Amp 660 CFM Rolling Dust Collector with 12-Gallon Bag and Optional Wall Mount , Black” and was $141.66. The other was described as “WEN DC3474 7.4-Amp Rolling Dust Collector with Induction Motor, 15-Gallon Bag and Optional Wall Mount , Black” and was $197.32.

The cheaper one had a brushed motor. The brushes are rated for 250 hours of use, which is only 10 days if you leave it running. I read up on what an “AC Induction motor” was, and it was inconclusive if it was a desirable motor because single phase ones are not as good as 3-phase units. To add more to my confusion, the cheaper one was listed as 660 CFM, and the more expensive one was called 600 CFM. I know that CFM ratings are unrelible even more than HP ratings, so there was no way to know without testing them both.

Here are my results with them working through about 10 feet of 4″ dust-collector hose:

Note how the Harbor Freight (with the filter bag) is not 10x the suction of a Shop Vac with it’s own filter. In fact, it is only about 91% of the suction of one of the smallest and lowest-cost models that Lowes sells – one that is on sale for just $62 now.

One very interesting thing is that the “7.4 amp” WEN lost very little performance when the bag was installed. It went from 495 to 489 CFM – almost 99% of the unrestricted performance. The Harbor Freight went from 441 to 252 – a much greater drop. Is this because the Harbor Freight bag is very restrictive, or is the blower motor so low on torque that it reduces performance dramatically when there is even a small amount of resistance? The answer is some of each, but mostly the bag. The Harbor Freight bag would not fit on the WEN outlet, but the WEN bag on the Harbor Freight did raise the performance a lot – indicating that the Harbor Freight bag is most of the problem. Whether that is due to it filtering to a smaller micron level, or due to it being less efficient – I don’t have the means to test. It is entirely likely it simply filters smaller particles.

These impellers all looked like they would clog and so a dust-separator seemed important. I have the Home Depot kind with 2.5 inch fittings, but much to my horror it wiped out the performance of all of these units, except for the Shop Vac! Turns out that these centripetal blowers are all very sensitive to flow restriction whereas a Shop Vac is not! I thought about building a separator with 4 inch fittings, but this was becoming a project, and I could just buy the Harbor Freight “2 HP” model for $50 more than the larger WEN, since the local store had one with a damaged box at a discount. I had avoided this unit earlier because it claimed to need 20 amps, but by now I was pretty sure that all of these specs were exaggerated, so I got it and tested it. Turns out it did draw a bunch of power when starting up, but once it was spinning, it only drew about 7 amps / 760 watts. Now this is interesting, as this unit actually does draw a full 1 HP (but they call it 2 HP). The suction was a bunch better than the others ones – 3x my Shop Vac, all while using less power. Yes, it was louder, and it did take about an hour to assemble – but it’s the only unit that fully satisfied me and the one that I will be keeping.

To summarize, the “1 HP” Harbor Freight had about 10% more performance than the lower cost WEN – at least when no bag was on either. The smaller Harbor Freight cost about 20% more than the cheaper WEN, but it has an induction motor, and the brushed motor of the lower-cost WEN is only rated for 250 hours of use before needing the brushes replaced. A “2 HP” Harbor Freight averages about 1 HP of power consumption, and likely makes about 4/5 of 1 HP of actual mechanical motion – and I sure wish they would just rate them truthfully. I felt like the suction of the larger Harbor Freight was the only one that was truly good enough, and that is my favorite – but it is massive and louder. Still, if I need something for portable use such as drywall sanding, I would just use the Shop Vac, so I think the larger one is the only one that makes sense for a workshop.

Buying a Chinese 6040 Router

After reaching the limits of what my Comgrow RoboCNC could mill, even with a 600 watt brushless motor upgrade, I decided to get a larger CNC router. Something that would have a chance of cutting aluminum.

eBay is saturated with beige 6040 routers for $1200 and up, but I zeroed in on the more expensive China-Zone brand for a few reasons. One is that they had a model with extra Z height (150mm), and I was pretty sure I would need that if I added a spoilboard and/or vise. Two is that they possibly would take extra care in parts and assembly as they had a brand. Three is that I liked the bare aluminum look over the painted brown look. Four is a guy on YouTube had a brown one and his Z axis was not rigid – this ChinaZone one does not have that issue. Maybe none of the other brown ones don’t now either, I don’t know. Five is ChinaZone has discrete stepper drivers, and I have seen others that have them integrated with the motion card, which makes it much harder to upgrade the motion card.

Once I knew that I wanted the extra Z, the 4th axis, 2200 watts, and the wireless pendent, I compared the various listings. They all had different seller names, prices, and delivery times. Turns out they are all sales reps for the same company. I placed an order for one that was $1750 and free shipping from NJ/CA.

Someone promptly wrote me back and said they were sold out of US inventory, and they could send me one from China but it would take 4-6 weeks. I declined, and then placed an order with another listing for seemingly the same item. They also wrote me – this time from What’s App, and also said out of stock in the US, but offered to send me one from China via Fedex within a few days for $200 more and no import duty. Knowing this machine is over 100 lbs and that I would save sales tax, that was a good enough deal to not have to wait. In fact my friend in China who ships machines to the US said he sometimes spends $600 to ship a 70kg package. I was suspicious as to why they didn’t show out of stock on eBay, but after they contacted me, I did see that they updated the listings to show shipping was from China, so it seems to be an honest mistake due to many salespeople sharing the same pool of inventory.

I said ok, and they asked me to send money via PayPal rather than eBay as they could only absorb some of the extra shipping cost if they didn’t have the eBay fees. I realize this would scare some people, but I could tell it was legit – I said ok, but asked them to send a PayPal invoice with a full list of what I was getting so that I would have purchase protection. They agreed. I had to help them write the invoice to my satisfaction (there is no purchase protection on industrial machinery, so I had to had it say it was a hobby engraver). I paid them, and got a tracking number from UPS. After a day, UPS kicked it back, and said the machine was too magnetic to fly (the spindle). They reshipped it by Fedex. After about a week, I had the machine.

First problem was they shipped a parallel cable rather than the USBCNC cable I requested. This brings up a bad choice I made. I assumed that the option that worked with USBCNC over USB would also work with Mach3 over USB, so I picked that card figuring I was getting a two-for-one. I was wrong. It works with (an old version of) USBCNC over USB but Mach3 *only* on parallel. I should have picked their option for Mach3 over USB. In any case, they shipped me a new cable via air, and it came a week later.

To avoid waiting, I bought some parallel to USB cables from Amazon, and none worked. So, I didn’t wait and got it working with the parallel cable and Mach3 (I later learned that LinuxCNC would have worked also). This was quite hard as I had to find an old PC and find a way to install 32 bit Windows 7 – plus I had to buy a $20 parallel port. Even then, it didn’t work at the documented profile settings. I had to learn on my own that PCI parallel cards use a custom memory address that differ from the instructions. Took a few days to figure out – how to find the address, etc.

I got it working, and many of the settings I had to use in Mach3 to get it working well differed from their instructions. Why they don’t just send everyone a working Mach3 profile, I will never know. Also, I eventually got the pendent working, but they never sent the driver to me, so I had to hunt that down. They could make this way easier and more turn-key. Seems like whenever they sent drivers, they were either not the latest one. As if no one there was maintaining a support package.

All seemed well at first, but sometimes steppers would lose steps. I noticed that the discrete stepper drivers were set to a rather low current setting. I turned them up to just over 3 amps, and when running, the steppers never got much over 40 degrees C. Steppers can easily run over 70C or more, so that seems good to me. This didn’t fix the problem though. I searched around, and found a post from 2012 where a guy said he had this issue and solved it by moving his 48v power supply out of the enclosure. I don’t think he needed to do this. I found the manual for the stepper drivers, and it said to use twisted-pair wire to help with interference. I changed my X and Y wires to twisted pair, and my problem virtually went away. But, I am not sure that it was the wire, because I also tightened many surprisingly loose terminal screws – so possibly the connection was being lost. Also, the parallel cable they gave me was not shielded. Amazon has $10 shielded ones, so I ordered one of those. Also, I ordered $10 in copper foil tape with conductive adhesive so that I could wrap all of the wires. The VFD cable appeared to be unshielded. It should be shielded, and also should only be grounded on the controller end. They did a really clean wire job in the controller, but didn’t do much related to EMI – although they did have choke magnets.

https://www.amazon.com/dp/B092QTB173?psc=1&ref=ppx_yo2ov_dt_b_product_details

https://www.amazon.com/dp/B071JKLFXX?psc=1&ref=ppx_yo2ov_dt_b_product_details

Meanwhile, this is a water-cooled spindle. I am glad that I have the 2200 watt one as even with this power, I still stalled it twice. It’s not as much power as you would think, or probably not even what they claim – so I wouldn’t suggest the 800 or 1500 watt option. Also, the smaller motors don’t use the collet size that can accept 1/2 inch.

It came with a spindle cooler pump, and you normally fill a bucket with water, and it circulates that. Problem is, the water would get very hot. I almost burned my hand – had to be over 125 degrees F. I ordered an “industrial chiller” for about $140 from eBay. It works great! The water never went over 35C after that. I should also note that I have owned two higher-end ~$400 laser chillers that had actual compressors and refrigerant in them. That is not needed for this. The $140 one is good, but you will need some tubing adaptors. I 3D printed mine, but Amazon has some brass ones. Get the chiller.

But, I ran into another problem. If I set Mach3 to 25kHz, it seemed reliable. At 35Khz, it was mostly, but not perfectly reliable (it would run for many hours). At 45khz, it was not reliable enough (would sometimes fail with 3 hours). So why not set it to 25khz? That was limiting my rapids to about 1200mm/min or about 40 inches/min. I decided that parallel port was a dead end, due to speed and also due to needing 32 bit Windows – and the USBCNC cable they provided me with did not work with any Mach3 plugin.

I looked into other motion controllers, and my conclusion is that the best would be either a UCCNC 300 or 400 Ethernet controller with breakout board (which would cost about $180 and work with both Mach3 and UCCNC software), or a $160 Mesa card, which would only work with LinuxCNC. I wasn’t sure I was up for configuring Linux on my one machine that I was using every day, and I didn’t want to wait for a shipment from Europe for UCCNC, so I got a $27 USB card from Amazon. That took me around an hour to install, and it worked. Only negative I saw is it only has 4 inputs. That is enough for XYZ limits and probe, but nothing is left for a 4th axis optical switch, or a water alarm switch. Not a big deal though. I am pretty happy with how it is working. Rapids are at about 2600mm/min now. Homing also works once you get the settings/limits figured out (hint – go into the plugin settings for the motion controller and you may have to adjust how much it moves backwards after it hits the limit switches). Also, get “limits” set up properly.

https://www.amazon.com/dp/B07PDYJBYX?psc=1&ref=ppx_yo2ov_dt_b_product_details

For Mach3, the UI is way too ugly to use, but this guy solved that with a nice screen set: https://www.youtube.com/watch?v=7o_V_1AIaXE&t=1173s

Now that it is all working well, I will say that the size is good as it fits brilliantly on a Harbor Freight cart that has room for the controller, the chiller, and a PC on the bottom. But, the few times I wanted to make signs, the size was marginal and I was fighting over every last mm. Whenever I am making signs, I wish that I had a OneFinity X-50. So why did’t I get a OneFinity? Partly because I ordered the 6040 before I knew about it, partly because the OneFinity is $3000 by the time you add a water-cooled spindle, and partly because there is a long wait to get a OneFinity. In any case, it would use up a lot more space. Why a OneFinity over the competitors from Carbide3D? It is vastly easier and quicker to assemble, and I like how it uses ballscrews over belts. Speaking of size, is it really a 6040? Between the limit switches, I came up with the largest practical soft-limit settings of 390×570 – so no. Ideally it would be nice if they designed the machine to truly have a cuttable area of 400×600.

https://www.harborfreight.com/tool-storage-organization/tool-storage/tool-carts/36-in-x-24-in-polypropylene-industrial-service-cart-58323.html

About enclosures. I was pretty sure I would need one, but I really don’t. I got a dust-shoe from Amazon that sometimes works great when connected to a Shop Vac through a HomeDepot dust separator, depending on what I am cutting. Also I run a Honeywell HEPA air filter in the room.

https://www.amazon.com/gp/product/B09VDK43TW/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

https://www.homedepot.com/p/Dustopper-High-Efficiency-Cyclonic-Dust-Separator-12-in-Dia-with-2-5-in-hose-36-in-long-with-2-Sweep-Elbows-HD12A/315749552

In summary, I am happy with the machine, but I spend many days fighting with it to get it going, and it all could have been avoided if they hired a truly native English speaker who knows about this stuff to clean up their documents. They spoke English nearly perfectly, but not enough to make unambiguous documentation.









Dog Dryer Comparison Test

I went through my entire past German Shepherd ownership not knowing that professional groomers save an enormous amount of time, effort, and stress on the dog by using a blower to remove undercoat. I spoke to a groomer, and they said they would never manually brush out a Samoyed – as the dog would get irritated with just how long it would take – plus their labor cost would be much higher. Now knowing that, and with a Samoyed puppy on the way, my research is telling me to get a “pet dryer” to blow away the loose hair.

It appears that the K9-II or III is probably the best, but it is $450, and that seems to make more sense for a professional than for home use for a single dog. I emailed them asking if I they could loan one so that I could include one in the test to see if it was actually better, but they didn’t reply. I did get to test five other units. The Homeend Dual Motor, the ShernBau “5.0” HP, the Flying Pig, the Air Force Commander, and the PetNF. I would also like to say that I am impressed that K9 does not list HP, because HP ratings on blowers and vacuum cleaners are 100% fraudulent. If anyone can find a unit that lists the correct HP, I will eat my words, but there is no such thing. So how do they come up with these fake HP ratings? From reading around, it seems roughly more or less to cool the unit to below freezing, and then measure the input power for the first 1/10 second of operation. They take this power surge, and convert the watts to HP. This is a con in my opinion, and if I were not libertarian, I would say it should be illegal.

There are two ways to ways to measure HP with various methods of legitimacy. The most correct way is to measure the motor’s actual outputpower. This is how engines are tested, but no one does it for blowers. The second way, less legitimate, but common in the carpet blower world, is to measure average power consumption under load, then convert watts to HP. The problem with this method is that some motors are 90% efficient, and some are 50% efficient – so it still doesn’t really tell you anything. In summary, ignore HP.

So how do we compare? We should use CFM – or cubic feet per minute of air moving. There is only one problem with that – of the manufactures that do report it, that also cannot be trusted. They will use tricks like measure it from the motor and not include the hose. For this test, I used an air-speed meter, a tube of a measured diameter, and ran each blower into this to see accurate and comparable CFMs. Further, I used a sound level meter to measure the C-weighted noise from one meter above each unit. Additionally, I used a Kill-A-Watt meter and measured watts of power consumption.


The Homeland Dual Motor was $148.95 at Amazon at the time of this review. It was the most powerful, at 159.0 CFM with both motors on. It had the most CFM per dollar, and so is a good value if you want the most powerful in this test. It has no manual heat control, but the air does get warm. It is not clear to me if it has heating elements, or if the motors make the air warm. Judging from the watts/CFM, it probably does have a heating element – either that, or it is very inefficient. I am not selecting this unit to buy, because I don’t want to blow hot air on my Samoyed in the summer when he is not wet. Even if it does not have a dedicated heating element, it still has a poor watts/CFM rating, so that extra wattage is turning to heat one way or the other. Also, it needs a 20 amp outlet to not blow a fuse if both motors are on. You can run one motor at a lower CFM to either be quieter for the dog, or to work on a 15 amp outlet, but still it might scare a puppy since the lower setting is still about as powerful as some of the other units on max power.

Use this link to buy if you want me to get a commission. Same cost to you:
https://amzn.to/3cUkmYG

Use this link to buy if you don’t want me to get a commission: https://www.amazon.com/dp/B07KP8PYH2/


The Shernbao, at $178 currently, is the one I am going to purchase. The CFM rating is close to the Homeland Dual motor, but it will work on a 15 amp outlet and is at least 40% more efficient – making me think that it might have an above tyical high-quality motor in it. Even with the heat on, it was 1670 watts at 118 volts, which is about 14 amps – so should be ok on most outlets. This is sold in a few colors. Mine is dark glossy purple.

Use this link to buy if you want me to get a commission. Same cost to you:
https://amzn.to/2NfzjJQ

Use this link to buy if you don’t want me to get a commission:
https://www.amazon.com/dp/B07K8P7PTY


The Flying Pig is $172, and I love this unit. The look of it is very satisfying to me, and it tied for the quietest unit. I like the lighter purple matte paint better than the darker glossy purple of the ShernBao, and the controls are very nice. So why am I getting the ShernBao over this? Because the CFM of this is significantly less and the price is about the same. I would buy this in a heartbeat for a short-hair dog though. I see why people like this unit.

Use this link to buy if you want me to get a commission. Same cost to you:
https://amzn.to/3d15Tdh

Use this link to buy if you don’t want me to get a commission:
https://www.amazon.com/dp/B00JV33WJ8


The AirForce Commander at $199.99 appeared to be a well-marketed product with lots of good reviews, but I would rate it last in my test. The hose is smaller diameter than other units, which will restrict airflow more. It is much louder than the Flying Pig while being about the same air flow. It has two speed settings, but no heat.

Use this link to buy if you want me to get a commission. Same cost to you:
https://amzn.to/2NcrFQL

Use this link to buy if you don’t want me to get a commission:
https://www.amazon.com/dp/B00063KHPE


The PetNF, at $98.97 was a fine unit, but like the Flying Pig and Air Force, not the best airflow performance for a Samoyed. Feature wise, it is the same noise level as the Flying Pig, with similar controls, and close to as much air-flow. I could see this as a good Flying Pig alternative for someone with a short-haired dog who wanted to save money. Since it uses more watts for less CFM than the Flying Pig, it probably has a less efficient and lower quality motor – that may or may not matter in terms of which would last longer, but I have no way to test for longevity.

Use this link to buy if you want me to get a commission. Same cost to you: https://amzn.to/2MQdvEQ

Use this link to buy if you don’t want me to get a commission: https://www.amazon.com/dp/B07PLG4275

In conclusion, it was hard to narrow down which was best for me, so I really can’t say what is best for anyone else – but at least you can see comparative noise, CFM, and wattage data.

Garmin Fenix 6X Sapphire vs iPhone 11 Pro Max with Polar H7 – GPS and Heart Rate

After telling a friend that I got a Garmin Fenix 6X, he said that they were no good for heart rate because the optical heart rate sensor could not capture peaks. This lead me to do some research, so I searched around and found a video – which seemed to support what he was saying:

https://www.youtube.com/watch?v=WvDfl_2zxmA&t=239s

The video showed that, compared to a chest-strap, the Garmin watch tended to not give readings about 165. Now I was a little concerned, so I set out to do a test. First I prepared my new as of September 2020 6X Sapphire that I got fresh from Garmin. Yes, this one has a black screen. I used a silicon band, and made it snug, but not overly tight. The GPS was set to GPS+GLONASS. Then I got my iphone 11 Pro Max, and paired a Polar H7 Bluetooth HR chest strap to it. I set this up with the Wahoo app, since Strava no longer supports direct connections to HR sensors. I headed to a high-school track so that I could also test the GPSs.

First, I walked 3/4 of the track to let the GPSs warm up for four minutes. When I got to the start line, I walked along it a few times, for the goal of being able to later make a segment from this start line. Then I circled around, and ran through the start line. When I got to the same finish line, I went back to walking. I then alternated running and walking for a few laps. In my mind, based on my experience, I got my heart rate past 170bpm.

Garmin 6X Sapphire GPS track of inner lane
iPhone 11 Pro Max GPS track of inner lane

As for the GPS, the Garmin watch was much better. I also saw this a few days ago when I rode a hilly and densely wooded MTB course with each.

Overlaid tracks – orange is Garmin, blue is iPhone.
Hear rate – blue is Polar H7 chest strap. Orange is Garmin 6X Sapphire.

The heart rate data was surprising. Mostly they tracked about the same, though the Garmin seemed delayed a bit. I am not sure what this means, since I started both tracks within 1/2 a second of each other. There is one area where the chest strap showed 213 bpm, while the optical watch never went over 180. Which to believe? Well, I am age 51, and have been using HR chest straps for many hours, including for stationary bike FTP tests. I don’t think I have ever seen a reading over 188 or so, so I tend to believe the 180 more than the 213. In any case, the Garmin watch with 11.75 firmware is capable of showing HRs over 165bpm. Also, I have been swimming with this a lot lately, and the HR appears to work well under water while the watch counts the laps.


In conclusion, I am loving this GPS accuracy, and the entire 6X experience. The battery life is great – between 5 and 16 days for me, depending on if everything like HR and pulse-ox is on or off. Compare that to my Apple Watch Series 5, which needed to be charged daily – and for which I could not wear at night since it was charging. I wanted to sleep-track, and with the Garmin, that is finally practical. The optical HR shows data that I suspect is of reasonable quality, and I am surprised that my long-time owned chest strap is showing questionable data.

RMRC Recruit / AR900 “Phönix” Build

After having experienced the fun of sending a Parott Disco on a long waypoint mission and then losing it in high winds, I wanted to build something even better – but use readily available components this time. If you want to make the same one, I suggest learning from various YouTube videos, such as Painless360 – and then see my notes here for specific areas that I had trouble with.

I used DJI FPV, and a Matek 765. While you can use an F405, the pinouts are different, so you won’t be able to use my notes.

Battery is 3S2P with Sanyo 18650 that are 3400mAh each and held by this frame: https://www.thingiverse.com/thing:4340601

Use Velcro to mount everything except for the camera mount, which I glued in place to eliminate wobble.

Make sure 765 is oriented with the plane icon on it pointing forward. If you pick another orientation, there is a way to configure it to work, but it is more steps.

ArduPilot Setup

You have to do the normal setup as in the docs – I am only mentioning unusual things that I had to change and that were not in the docs. Here are some of those trouble spots:

Airspeed: You can install like my photo, or in the wing. You do not need to connect both tubes. Just connect to the straight connector and cut off the angled one. Also, it doesn’t matter which inlet you connect the tube to on the sensor as either will work because it just changes to the one that it sees more pressure on.

In Mission Planner, go to Setup->Optional Hardware->Airspeed

Select “Use Airspeed”
Ignore Pin for now.

Set Type to Analog

Now in Mission Planner, go to Config-> Full Parameter Tree

Change ARSPD_PIN to 10

Click on Write Params on the right side of the app.


In DJI googles, under Servo Setup, set AIL to reverse. All others stay not-reversed.

In Mission Planner, Setup->Mandatory Hardware->Servo Output
Set function in this order: Throttle, Disabled, Elevon Left, Elevon Right. The rest are disabled. Elevon Right is reversed.

Go into Config->Full Parameter Tree->RCMAP
Pitch (2), Roll (1), Throttle (3), Yaw (4)






Matek 765 connections:

Airspeed sensor uses Sig (yellow), 5v (red), and GND (black). These connect on the F765 to the area to the right of the SD slot where it says 5v, AIS, and GND.


GPS 880 Connections:

Yellow SDA on GPS goes to DA1 on the F765
Black GND on the GPS goes to one of the G
White TX on the GPS goes to RX2on the F765
Green RX on the GPS goes to TX2 on the F765
Red VCC on the GPS goes to 4v5 on the F765

Gray SCL on the GPS goes to CL1 on the F765

DJI FPV air unit:

You will have to extend the wires – I used 26 gauge servo wire and solder and heat-shrink tubing.

Red Power goes to the 9V pin on the F765
Black Power GND goes to one of the G pins on the far right edge of the board.
White UART RX goes to TX1 on the F765

Gray UART TX goes to RX1 on the F765

Brown Signal GND goes to one of the G next to the 9v pin
Yellow DJI HDL Signal goes to the RX6 pin on the F765

3DRV2 telemetry radio wiring:

Pin 2 on the 3DR goes to TX7 on the F765
Pin 3 on the 3DR goes to RX7 on the F765
Pin 1 on the 3DR goes to 5V on the F765
Pin 6 on the 3DR goes to G on the F765

RTS and CTS are not used.


Servo connections:

The throttle “servo” wire goes to S1, Vx, and G (the first servo connection)
The left elevon servo goes to the S3 servo position.
The right elevon servo goes to the S4 servo position.

Cybertruck Solar Roof – 15 Miles a Day?

At the Cybertruck announcement, Elon said that there may be a solar-roof option that could add “up to 15 miles per day” of range. This study is one quick check to see if that is possible.

First, we need to decide what “up to” means. According to the Guinness Book of World Records, Yuma Arizona is the sunniest city on Earth with 11 hours of sunlight each day. Elon doesn’t tend to make claims that are only obtainable in rare conditions, so for the purposes of this study, I will pick Los Angeles, CA – as it both gets a lot of sunlight and has a lot of Teslas. Further, rather than go by “daylight hours,” we need to use “peak solar hours,” as they a standard used to calculate PV cell output. This chart lists 5.62 per day on average for LA.

As per standard solar calculations, each solar hour results in 1000 Watts per square meter of light energy hitting the PV cells. While most home solar modules are 15-20 percent efficient, the cells could be 22% efficient – especially by the time this is available – so that is 220 watts per square meter per hour, or 1.236 kWh per square meter.

Now we need to know the size of the solar array. The best source that I could find was this CAD drawing. According to scale and from making my own 3D models using measurements from it, the door would be 1.3 meters wide. The length would come out to 1.8 meters, or about 1.815 meters factoring in a 7.5 degree angle for the roof. For the purposes of this study, we will assume no gaps in the cells to account for the way that the cover has slats with hinges, so the area is therefore 2.34 meters.

Now for the top portion. It is 1.3 meters wide at the top, 1.5 meters wide at the bottom, and 1.26mm long, including factoring in the 7.5 degree angle. That is 1.764 square meters. So combined, we have 4.1 square meters of space. I am going right up to the edge, and that is not really possible in the real world, so we could easily call this 4 square meters – but let’s give every benefit of the doubt and stick with 4.1 – and we already skipped counting gaps between roller-shutter slats as well. Now we can say that 4.1 square meters times 1.236 kWh per square meter is 5.07 kWh per day for an average day in LA.

How many miles per day is that? Well, a Model-3 Performance that has a 75 kWh battery and a 310 mile EPA range goes 4.13 miles per kWh – so it could run for 21 miles on that. But how much more energy would a Cybertruck use over a Model-3? It is not enough to just compare coefficients of drag because you also need to factor in the frontal area and tire rolling resistance. George Bower did a detailed power analysis and believes that the tri-motor truck will have a 250 kWh battery. For a 500 mile EPA range, that is 2.0 miles per kWh.

5.07 * 2.0 is 10.14 EPA miles per day in LA, a rather nice solar environment. So, it looks like a solar roof that is both on the roller door and roof-top would provide up to 10 EPA miles per day of range. A peak summer day in AZ is 7.9 solar hours, and could hit 14-15 EPA miles. So, Elon’s claim of “up to 15 miles per day” is possible for a bright summer day in Arizona, but expect more like 10 miles per average day in CA. Boston, by the way, would be 6.9 EPA miles on average – and 5.4 in the winter. So not bad! Note that these calculations assume that you are facing North and have no shadows on the vehicle. If you face South or have shadows, then there would be less sun.

Also I wanted to mention that I sell Tesla J1172 charge-lock adaptors that are SLA resin printed and then painted black for UV resistance. While not injection molded, they are a lot nicer than parts printed on home-level FDM (filament) machines.



The $45 Turbo Levo Hardwired Light

I didn’t realize sooner that there are ways to hardwire a light into a Turbo Levo because because there is no USB plug or other external power connector on the bike, but it turns out that there is one under the motor cover.

Watch the above video for visual instructions, but in a nutshell, I bought a Lupine cable for Brose motors, and connected one end to the inside connector, and the other end was fed out of a cable pass-through near the handlebars. To snake the cable, remove the non-drive side crank with an 8mm hex-key by turning it counter-clockwise, then remove the motor cover’s four screws. Inside on the bottom of the motor, there is a small round cap. Remove it with needle-nose pliers to expose the tiny power socket. Now, remove the bike control unit/display (on the top tube of 2019 and 2020 models), and use a snake-wire to push it through to the area behind the rear shock that has the brake and shift cables. Finally, snake the cable a second time down to the motor and plug it in

This is a non-referral-link to the light that I used from Amazon. It claims to be 5000 lumens, but in reality, it is almost as bright as my Gemini Duo 1500 when that is on medium setting. I measured 13 watts of power draw (1.08 amps at 12.1 volts). Whether this means 700 lumens or 1300 lumens, I am not sure.

You can cut the jack from the battery pack that it comes with and push that cable into the cable pass-through that is on the left or right side of the frame near the handlebars and then solder it to the Lupine cable wires. This is a nice connector because it is threaded for security and has an o-ring seal. I actually wanted to save this battery pack for other uses such as helmet mounting when I ride another bike, so I opted to instead use a 12v jack that I already had. In actuality, I may never use it this way because I own a Gemini Duo light for my helmet, but who knows – a spare battery is useful to keep in the backpack for emergencies. I did test the battery for 2.5 hours driving my Gemini Duo 1500 on low power before ending that experiment and deciding the battery that it came with was surprisingly not junk.

Note that you can’t use just any LED light head on a Turbo Levo due to the heads wanting 7.2 volts and the bike putting out 12v. The light head that I picked has so far been ok running off 12v, but I tried a different kind and burned it out in under a minute. If you want to use a light head that doesn’t do it’s own voltage conversion, then perhaps use these voltage converters

So why this light and not the dozens of others? I own a few Nightrider lights, and like them a lot. I also like my Cyglolite and Gemini. But, the Brose motor output is limited in wattage, and my existing Gemini and other high-powered light heads would not work. I really needed something with less wattage, so decided to try something that would have a better chance of working.

I actually purchased four different lights somewhat similar to this one and tested them all in a dark room. Two of the other ones had three LED emitters rather than two. Even so, this double-LED light was a bunch brighter than the other ones, as well as smaller and lighter – so it wins. For a test, I just did a 10-mile trail ride with a group of guys and two of them had their lights die on the ride and had to stop early. I, of course, had no concern of that happening and it worked out very well. Also, I like how it has no strobe mode to have to cycle through when changing brightness levels.

I then tried *two* of these units at the same time using a Y-adaptor. It worked! Power draw was 1.95 amps at 11.5v, which is 22.4 watts. Fantastic! For some reason, these heads stay within safe limits without a startup surge that is making other light heads go over the limit. I am going to be conservative though and say that I am not sure this is any more than a real 1400 ANSI lumens total. Total cost with the cable is still under $70 USD, and that includes two battery packs that I can use for my helmet light or for a manual bike.


Alternatively, there are very nice German-made lights specifically marketed for eMTB. One such light is the $295 Lupine for Brose motors, and another option is the 299 Supernova line of M99 Mini lights. Just note that, as previously mentioned, some of these lights exceed the maximum wattage capability of the current Brose motor, so don’t rely on their websites and double-check with an email or phone call with those companies as to which light is the most likely to work.

Finally, you may enjoy a 36-38T chainguide adapter or DeSlackinators for SRAM brakes.

.

Can an eMTB be as much rider effort as a MTB?

Sometimes I want to ride on my eMTB with people who are on normal/manual/clockwork/acoustic/old-school bikes and want to experience the same effort that they have while still having the ADA OPDMD ability of the eMTB to bring me out of the woods quickly if needed, so I set out to see if there was a small amount of assist that would make up for any motor drag and added weight without giving extra boost. I have seen people suggest 10%, 15%, or even 5% to make up for the weight and motor drag of the eMTB. My theory was that about 10% assist was needed to match the effort, so I devised an experiment to find out. I will define success as coming within 1 mph of the same average speed for the segment with the same amount of effort, as defined by average heart rate.

My manual bike is a 2013 Giant Trance X 0 29er with a Stages power meter and is approximately 31lbs. Mods include Easton Arc 30 rims with tubeless Nobby Nic Addix rear and Maxxis DHF MaxxTerra EXO+ front – both 2.6 inches. My eMTB is a 2020 Specialized Turbo Levo and is about 50lbs. Mods include 35mm internal rims, Nobby Nic Addix rear and Magic Mary – both 2.6, DeSlackinators for the brakes, and a 38T Chainring . Tire pressure was about 15psi front and 18psi rear for both, which is appropriate given the 2.6″ 29er tires, and my weight of about 155lbs with full gear, including backpack.

I rode the same Strava segment, named “VC-Turkey Hill Loop,” with each – matching my heart rate the best that I could. This segment is a mix of single-track through forest, some paved bike paths, some gravel roads, and some fire roads.

First, the manual bike. My heart rate was 151 bpm average, and my Stages power meter reported 172 watts average. I completed the 7-mile loop in 36:13, for an average speed of 11.6mph.

For the eMTB, my heart-rate was an effectively identical 152 bpm average using the same Polar chest strap, and the Turbo Levo’s pedal force sensor reported 156 watts average. I completed the same 7-mile loop in 33:39, for an average speed of 12.5 mph. Assist was on 10%, with the “peak power” setting at 49%. If the heart-rates were the same, why was the power input to the pedals different? Based on my 1000+ km of riding the Turbo Levo and seeing this issue time and time again, I believe that the Turbo Levo power meter under-estimates power compared to my Stages meter and appears to do so by approximately 9.5%. If there is a way to calibrate it, I would like to know how, but it is to be expected that no two power meters will give the same results, unless carefully calibrated to match beforehand. I will make use the heart rate to say that the efforts were equal, since heart rate data from a chest strap is very reliable.

As we can see, 12.5mph of the eMTB on 10% assist with peak power on 49%, is faster than 11.6 mph of the manual trail bike, so setting a Turbo Levo to 10% assist is a modest-amount easier than riding a manual trail bike for this course. Perhaps a value of 8% would be a more exact match to the manual bike, but then again, this Giant is not as fast as a cross-country bike with faster rolling tires, like Rocket Ron, Racing Ralph, or Racing Ray – so 10% is on the high side, but probably fair enough compared to a more optimal manual bike with faster tires.




How to Prolong eMTB Battery Lifespan

eMTB batteries are expensive – some are about $1000, so it makes sense to want them to last for as many charge cycles as possible. The two things that you can do to prolong your battery life is to not fully discharge it, and also, perhaps surprisingly, not store it fully charged. For example, this article shows that if you discharge to 40%, their example cell-phone battery will last 1500 cycles until it has 70% of its original capacity still functional. But if you discharge to 0%, it is only 600 cycles. They give further examples of how if you leave the battery in their example fully charged for a year at a 25c temp, it will lose much more of it’s original design capacity than if stored at a lower percent state of charge.

My 2020 Turbo Levo with the 700Wh battery can last a long time, and so I usually only need to charge it every three rides. For example, a typical ride may use 33% of it’s capacity. Charging it every ride to 100% would be bad, as seen on the charts, because that would result it in being stored 100% charged. And only charging it when it completely dies would also be bad, as seen in the other chart from the linked article. So to make the battery last as long as possible, I should ride until it is at 20-40%, and then only charge it to 60-80%. If a longer ride is expected – it is no problem to charge it to 100% before the ride, because it is not being stored at that charge for a long period of time, so for sure, charge to 100% when you need to.


The Giant eMTB has a charger with a button specifically designed for this safer storage – push the button and it will charge to 60%. But what should a Turbo Levo owner like me do? You can buy a timer like this Stanley from Amazon. When using it, my 700Wh battery went from 42% to 64% in one hour – so, 22% gain per hour. This means that if I do a ride, and my battery is at 40% and I want to charge it to about 80%, then I need should push the “2 hour” button on the timer.

Another way to look at is is that a 2-hour charge will cover any of my normal rides, so I just push the 2-hour button most of the time. If I know I want a full charge, then the 6-hour button will always cover it. If you want to know more, read the linked article, or countless others. Or just only fully charge the battery before you need to use it at full capacity and try not to run it too close to empty.

Winter Wheel Set for Turbo Levo

My previous article on an attempt of finding the best tires for my 2020 Turbo Levo only reinforced that there is no one best tire so much as the best tire for the current trail condition, so, I decided to get a second wheel set for the bike so that I could alternate between tire types – especially for winter riding.

I have come to prefer riding a normal MTB in the winter vs a fat bike because I had no luck with riding a fat bike unless the snow was thin and hard-packed. Even then, if I hit a patch of ice, it was super scary. The solution was carbide-studded tires, which are very expensive for a fat bike, but only about $150 a set for a normal MTB if you shop around. I have the Ice Spiker Pro, and like them a lot. They are even great in wet weather and grip wet wooden bridges like nothing else, though they are very loud on pavement.

The 2019 and 2020 Turbo Levo needs 15×110 thru-axle up front and 12×148 in the back with a Shimano driver for the base or Comp and an XD driver for Expert or S-Works. Also don’t get Centerlock brakes as then the wheel sensor magnet won’t mount.

There are various wheel sets on eBay in the $280 to $600 range, but I decided to give my local bike store a chance, and Rockland Cycle in MA offered me a 30 or 35mm internal width 29er wheel set assembled in Florida USA by Wheelmaster (part number 742107) using Ryde Edge M35 rims, DT Swiss 2.0mm black stainless spokes (32 per wheel), and Origin8 MT3100 hubs (36 pawl engagement vs 20 for the stock Specialized hub). It would not come with tubeless tape, valves, rotors, or a cassette, but he said he would put tape on it at no extra charge. I said ok to $269.99 plus tax, and he called me the next day saying they were ready to pick up.

Rim specs

Weight with the tape but no rotors or cassette was 1134 grams front and 1344 back, which is expected since these are rims specifically sold for eMTB, Downhill, Enduro, and Free Ride. Compare this to 1020 grams for the stock front and 1222 for the stock rear. Note that the same wheels with 30mm inner width would have saved 115 grams. So, for the same width, these combined are 120 grams (4.25oz) heavier than the stock wheels. For comparison, a $600-$900 wheel set with double-butted spokes and alloy nipples would be almost 1 lb lighter, front and back combined. Is it worth paying $300 more to save 300 grams? That is $1 per gram. Everyone can decide for themselves, but for me, that makes more sense on a Triathlon bike than an eMTB – especially since I just wanted winter wheels. But, you could get a high-end upgrade wheel set for normal use and keep the stock ones for winter.

As for 30 vs 35mm, my personal opinion is that most people should get 30mm, but a recent MBTR article said “Many observers believe the majority of riders of all stripes will settle on 2.4 to 2.6 tires fitted to wheels with internal rim widths of 30mm or 35mm, with a lean toward the latter. The Ibis 942 and 742 (35mm internal) rims, for example, outsell their narrower 29mm cousins by 9 to 1.” So, maybe 35mm is in more demand, and there are lots of tires coming out designed for wider rims.


I mounted my Maxxis Assegai and DHR2 tires easily and they inflated without tubes or sealant using just a manual floor pump. After they inflated, I let the air out and added 100ml sealant into each one using a syringe.

For rotors, I cheaped out and got $9 each ones from Amazon, and they are fine. Laser-cut steel is laser cut steel. 200mm ones are needed front and back. Problem is, these were 203mm, and that was enough to make them not fit. I used #10 stainless washers, two under each bolt, to raise the calipers, and that is working well.

For the cassette, I decided not to get another SRAM NX as there are much lighter ones for the same or less money – both the Shimano M8000 11-42 and the Sunrace MX8 seemed better, and either of those looked good to me. I ended up with the Shimano, and it is 435 grams. The stock NX 11-42 is 527 grams or so, so the change was within 1 oz of making up for the difference in wheel weight. You will have to either move the speed sensor magnet, or get a second one.

Once everything was mounted, I did a wet-weather run with the Assegai tire, and then the same route with the Ice Spiker Pro. The Assegai was great on the trail in general, but I could make it slip by testing panic stops on wet wooden bridges. Not so with the Ice Spiker – it was like Velcro ® brand hook-and-loop fasteners even on wet bridges. Shifting stayed indexed on both cassettes, and after some adjustment, either rotor worked with no rub. The same brake lever Deslackinators worked fine as well. I will warn that had I tried to go to a 46-tooth cassette, the chain might not be long enough to continue to work with my 38 tooth chainring.

In summary, these wheels are good for the price, but heavier than wheels that cost 2-3x as much. You can probably get a stock wheel set as a takeoff from someone for the same price, but it is unclear if that is better or worse – especially since this rear hub has more engagement points. They both will have 2.0mm non-butted spokes and brass nipples, and I don’t know which which hubs will last longer. If, however, you want wheels that are an actual upgrade, you should look for something with lighter hubs, and rims made from 6069 alloy. In doing so, it is easy to spend $600 to $900 for better alloy wheels that save about 1lb of combined weight, or you can just lose 1lb of weight off your body.