PWMing the power supply to the fan is what we’re talking about here.
1 Like
I thought so, just pointing out that if he thought he had a PWM fan working in the past that that is not the same thing.
2 Likes
For sure. That’s not really the situation here, though.
JP1 has a switched low-side output similar to what’s on the board here. That would work even less well for a 5V signal PWM without some pretty meaty external pull-ups due to the FET output capacitance.
Bart just posted a new ESP32 S3 board design. He has (4) 5V Digital/PWM outputs. I thought this might help by comparing his options to the JPv3 even though his has the S3.
3 Likes
Apologies for the delay. Back at it.
I found I had mixed up the 5V outputs of gpio.0 and 2, made me take a break yesterday.
I understand the scope issue now. If I am trying out the 24V socket in any way I can only use that negative since it is lowside switched. Hooking up the USB or the board ground ties that all together…not good. So I am not worried about this anymore.
The PWM outputs look fantastic. Using the analog_output instead of the Laser output made this much easier. I am getting perfect signals, 100khz just like the JP1 1% duty cycle actually look usable, but definitely at 3%. Using analog outputs is much easier to test with instead of using the laser output map stuff, saving me from double converting. Needless to say 5000hz works perfectly in 5V.
It does actually seem to work pretty well on the 24V side as well. I am not sure how to market the capabilities for that though. The key here is reading through the fluid NC wiki I see that servos like ~50hz, and I know fans will not run at low percentages (PWMing the power at least). So I can control fans 30% - 100% just fine. I have nothing else to test with since the scope does not like measuring this low side switching (although I can see it it is just rough). So in the end the 24V side seems to be capable of a lower frequency PWM it still does work.
I tested with the 24v output caps in the right place, all good.
So I messed up the position of the 24v output caps and swapped the 0 and 2, 5V outputs. Otherwise, I think this is 100%.
3 Likes
Will the Jackpot3 be available through Elecrow as well?
Together with the international version of the hardware kit that would spare me all of the rather expensive US shipping.
Absolutely, I already have my deposit waiting…but I will not place an order until after a few more people try these out, and then they will have to get made. So probably a 4-7 weeks for elecrow, best case 3 weeks for US boards if all goes well.
3 Likes
You could hook some high power resistor, a heating cartridge or a heated bed up to the output and measure from the switched ground to actual ground. That would eliminate all those funny ground loops.
With a few watts of resistive load, the rise time should be not too badly distorted I think.
1 Like
I’ve not sure what the relation between the 24V output and servos are.
As long as the 5V output can do 50hz with pulse widths of 0.5ms to 2.5ms you should be good for 180 degree PWM controlled servos.
1 Like
I just had no idea it was useful so slow.
I need to try that, thanks!
1 Like
Do you mean for the scope? I’d still say you shouldn’t be clipping the ground probe to anything that isn’t a DC voltage on an isolated circuit and definitely not anything other than ground for circuits that may not be isolated.
Nice! As with all troubleshooting, it’s just a case of working through the process 
It comes down to what you’re doing with the PWM and how it reacts. Anything inductive like a motor or creating physical motion/heat will typically be slow. Same thing with most lighting for human-visible stuff, less so for machine vision.
Pretty sure I suggested that twice previously 
And do try using the math channel to re-create the waveform the 24V load is seeing by probing the +24V and the switched node… That’s a useful skill to have.
3 Likes
I was missing the load part of that. I messed with it for a second when you said it but then when I didn’t see anything I moved on. I see it now. I also the math thing is cool!
So looking at that. The 24V PWM looks good, actually at all speeds.
So I am understanding this stuff more. The fan just really does not like high PWM, good info to know. I never even considered it. But it makes complete sense.
As for the port location mix up…Jono you were right. It was my weird schematic, I lost track.
I feel so much better now that I do not feel completely lost. Once again I learned something, dammed if it does not feel amazing to learn something like this. The downside to learning more is just realizing how much more I don’t know. Odd that it is so much more clear today than yesterday.
As always thank you everyone for helping me out.
4 Likes
Yeah, it’s a useful thing to have in your toolbelt, for sure. It also lets you do stuff like plot realtime power draw if you’ve got a current probe or a scope probe you can place across a current shunt (although care is still needed here due to the shared ground situation).
Great result As Wittiko said, the edges will look cleaner the more loaded they are, up to a point.
To expand on this, it’s because the fan essentially has its own controller to handle the brushless DC nature of the motor. It turns on, senses where the magnetic poles are at switches power to the coils accordingly. When you’re controlling it via the PWM pin you’re telling that controller how fast you want it running and it’ll aim for that target itself. When you’re controlling it via PWMing the power supply, you’re basically turning it on for say 10ms during which time it starts up, gets synced to the fan rotation speed and then starts speeding itself up towards its max speed target. You then turn it off for say 10ms in which time it starts to slow down again naturally before turning it back on again etc. That’s one of the reasons why you’ll sometimes get better results starting at 100% for a second or two and then slowing down to 20% rather than trying to turn on at 20%. It’s also why it stops working if the PWM speed is too high, it won’t be on for long enough to get through its startup/sync routines etc.
Yeah, we’ve all been there. Most of the advice I’ve given is based on things I’ve done as a way to prevent mistakes I’ve made or seen others make previously.
Awesome. Just a heads up, next design you do that doesn’t work you’ll likely feel like you do currently. Every time I have a design that doesn’t work and I can’t figure it out I have to tamp down that impostor syndrome feeling and just remember that I figured it out eventually last time. Then when I do find it, plenty of times that has led to colleagues being stunned that I was able to even find that issue or decipher what was going on, despite it feeling not that bad in retrospect, etc.
It’s a process. Glad it worked out
1 Like
Yeah this time was like a triple hit, seemed like the computer was busted, the scope, and the board. I was spinning trying to figure out how that could possibly happen, instead of the obvious it was one, not all of them…and it was not even one of them it was operator error on the scope. I was so flustered, I was repeatedly testing every output socket and repeatedly wondering why some were not triggering, so I just kept turning them all on or moving the plug around to figure out where my signal was. I just kept working around it and never thought to look deeper since there were bigger fish to fry.
In the end this is definitely better than the jp1, and even better than the jp2, connectors are awesome (even though I hate the extra ordering), and we got a bonus output, fixed the gpio.26 blip on power on (we have a blip on gpio.0 but that was to be expected).
I am very excited about this board but for the end user, it is not all that different from the jp1. Odd how hard incremental improvements are.
3 Likes
Eeeeexcellent. This feels like a personal victory for me. Vindication! I should really send you a photo of some of my other control boards that are basically size constrained by how many of those connectors I can fit around them!
Yeah, although in this case I think it’s down to doing more stuff yourself rather than lifting stuff from others. Not that either is a bad thing, but it’s more work to DIY than to copy, and then more work again to do something complex yourself etc. The more you learn, the more ambitious you are and the more aspects of the design you’ll evaluate, etc. etc. At first you were just stoked to grab a schematic chunk from someone else and have it work. Next time you’ll be doing simulations to check edge rates and wanting things to be a bit faster. I think you’re now in a better position to go back to considering high-side switching, etc. for instance.
It’s weird, but all of this never ends up feeling ‘easy’, I guess. It’s really just that I’m no longer even really aware of the easy parts and I instead focus on the stuff that’s hard for me. Some of that will be easy for other people, some of the stuff I don’t even think about is hard for other people. Fundamentally I’m employed doing stuff that there are maybe 100-1000 people in the world working on and familiar with, reminding myself of that helps. The number of people who have designed, produced and sold a CNC controller is a pretty small group, too.
3 Likes
Included drivers and ESP are so nice. No more chance for me to fry my genuine V1 ESP. 
It really helps the out of the box experience.
2 Likes
That is really the goal. I know this is a DIY thing, but building a machine and assembling all the components into your own board seemed like a bit too much, so then I assembled them. This should get us on the right path for a while.
I would love to work on an S3 board but honestly I don’t see a need for it right now. We have lots of IO for what we do and unfortunately this one is ready to go and S3 support is still in beta from FluidNC so no point is swapping out processors. So I am going to keep moving forward with the J3 for a while.
6 Likes
Slowly working on my first LR4 build, looking forward to starting off my journey with a J3!
5 Likes
We have a WINNER! The JP3 worked perfectly on my MPCNC with a laser. Great job getting this all figured out Ryan! Also, thank you to all those that helped Ryan along the way. What a great community we have here at V1. 
Here is an example from the Jackpot v1:
Here is the Jackpot v2. You can see that it was cutting at 100% the whole time.
And here is the Jackpot v3. 
11 Likes
Sweet, we have a winner!!! I love the back to back same file. Amazing.
Like these plugs better?
Anything that could be changed?
5 Likes