PWM Dimming on my VFD clock

More
2 months 4 weeks ago #11472 by TexasPenguin
Hi All,
I've built a couple of "Ian knock-off" Nixie clocks in the past (with a little help from Ian I would add) and then I moved on to VFD clocks.

I took various designs and came up with something that worked well.
Critical to the design was auto-dimming, which I did by using a high side Moseft switch to PWM the 30V supply to IV-11 tubes driven by UDN2981 drivers.

Well, all was good for my first 3 clocks but then I ran out of the discontinued UDN2981 SOIC chips.
So, I found on Ebay some UDN2982 devices which are allegedly identical apart from higher voltage capability.

The issue I have is that, when PWM dimming, the OFF outputs are pulsing at the PWM frequency, giving partially on segments. It's definitely the UDN devices since I lifted a pin to isolate from the VFD and sure enough it is pulsing.

My immediate thought was - cheap Ebay devices, probably fake.
But then I scoped one of my original clocks and saw the same PWM pulse on OFF segments, but just much lower level so as not be very visible on the VFD segment.

Anyone got any ideas, I use a good amount of decoupling and I can't find any suggestion that you can't use a PWM supply with these drivers..?

I have ordered some more supposedly UDN2981-compatible devices from Farnell but my nagging suspicion is that there's a fundamental design flaw that's going to bite me.

Attached is a scope pic showing (top trace) and ON segment and (bottom trace) an OFF segment.
Schematics are also attached.

This browser does not support PDFs. Please download the PDF to view it: Download PDF

Attachments:

Please Log in or Create an account to join the conversation.

More
2 months 4 weeks ago #11473 by Ty_Eeberfest
Am I reading that scope right? It looks like your PWM frequency is 31.37KHz. Or am I misunderstanding?

Data sheet says the 2982 can have a turn-off delay of as much as 10uS. At 31KHz that is problematic. Why do you want to PWM that fast? 4KHz or so is more than fast enough for this application.

According to my sources 2982 is as obsolete as 2981. You could have used MAX6921 and eliminated the 74HC595s as well as the 2981s. But it's probably too late for that?

I've had to deal with the 2982's low switching step-cousin, the ULN2803 before. I found them to be unreasonably sensitive to tiny currents including noise on their inputs. The internal transistors get into their active region and you get "partially on" outputs. They act more like a current amplifier than a switch. I was not impressed and won't use them again.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

More
2 months 4 weeks ago #11474 by TexasPenguin
Thanks for the quick review, Ty.
Yeah, you are reading the scope picture right, 31.372KHz.
The problem with 4K is....it's 4K, and you can hear it - and it's pretty loud.
And, in any case, I'm not really concerned if the outputs don't switch off fast enough since the min PWM m/s is programmable and all I'm interested in is apparent VFD brightness.
And, in any case(2), even when I tried 4K and even lower, the PWM is still there on the OFF outputs.
And yes, designing in the MAX6921's would have been a great idea - before laying out the PCB's a year ago!

But your thoughts about input sensitivity may have some merit, maybe I'll look into filtering on that.

Thanks again...

Please Log in or Create an account to join the conversation.

More
2 months 4 weeks ago #11475 by Ty_Eeberfest
One possibility is that the PWM signal is rising so sharply that it's causing some sort of dV/dt weirdness in the transistors of the 2982s. Try a little capacitance, maybe 10nF or so, from drain of the IRFD9024 to ground. Check the scope and see what happens. I'm not sure 10nF is the right value, it just seems like a good place to start. You don't want to turn the PWM into a triangle wave, just lean the edges over a tiny bit. Seems far fetched but it's easy to test and just might give a better idea where the problem actually is. It could be that the 2982s don't like sharply pulsating input power.

I wonder what is vibrating to make the 4KHz audible.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

More
2 months 4 weeks ago #11476 by TexasPenguin
Yeah, tried capacitance on the 30V supply, that was my first "fix". 100nF made little difference, 4.7uF (what I had) turned the PWM into a triangle and did solve the visual issue but gave me inrush issues when turning on. Plus, it just didn't seem the"right" solution.
Since the 4KHz sound (and lower frequencies) was so loud and seemed to come from everywhere I just assumed it was the grids themselves vibrating.

Please Log in or Create an account to join the conversation.

More
2 months 4 weeks ago #11477 by Ty_Eeberfest
I don't have any other ideas right now. Just out of curiosity, when you tried the 4.7uF was the now-triangular PWM still showing up at all on outputs that were supposed to be off? I'm still trying to decide if the problem is some kind of crud on the inputs to the 2981s or if the 2981s are unhappy about their power input being pulsed.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

Moderators: AccutronTy_EeberfestIan
Time to create page: 0.100 seconds

Search

Tube Suppliers

Go to top
JSN Boot template designed by JoomlaShine.com