Needing more than a spark test?

[rabler]

On a NUC or any proper PC or Mac, Arduino IDE2 or 1 is ok.

Problem is that I'm currently using an RPI 4 w/ 4Gb as my host for the RP2040 MCU using the Pico SDK, which works fine. The easiest would be to install the Arduino IDE on there, rather than swapping computers around. I'll probably try it on the RPI 4 to see if I can live with it. I have a linux laptop that I could use for the Arduino IDE, but my shop desk is getting crowded with three computers (counting the RPI 4). Not a show stopper, just an organizational issue.

 

[WobblyHand]

Problem is that I'm currently using an RPI 4 w/ 4Gb as my host for the RP2040 MCU using the Pico SDK, which works fine. The easiest would be to install the Arduino IDE on there, rather than swapping computers around. I'll probably try it on the RPI 4 to see if I can live with it. I have a linux laptop that I could use for the Arduino IDE, but my shop desk is getting crowded with three computers (counting the RPI 4). Not a show stopper, just an organizational issue.

It does work. It is slow (a lot slower than a PC) but it does work. That's what I have to use to program my Teensy, because my cranky laptop dies if I program a Teensy. IDE2 is much nicer than IDE1, it's more like a real IDE. I use IDE 2.0.4 on an RPI4, so I know it well.

Arduino is not supporting IDE2 on RPI actually no generic Arm32 or Arm64 devices. But they do support Mac & PC. You cannot download IDE2 for an RPI from the foundation.

I found the appImage at https://github.com/koendv/arduino-ide-raspberrypi . I am using his build. Only supports 64 bit OS, which I installed on my RPI4-8GB, just for this. It runs, I use it. Performance is tolerable if running off a USB3 SSD. I wouldn't dare execute the IDE2 from SD card.

 

[homebrewed]

I have three RPI-like boards from hardkernel.com. Two are C1's, not currently offered -- one is running as a call blocker using NCID and the other as a print server. The third is an XU4, used as a media server to drive a 4K HDTV. Not quite as cheap as a current-vintage RPI. Their benchmarks look pretty good compared to a PI3, info here. They don't seem to have a problem keeping them in stock.

 

[homebrewed]

Notes on using my CCFL driver board(s) as the basis for a PMT power supply. Loading the HV output with a capacitor that was supposed to simulate a CCFL lamp didn't really drop the peak voltage output all that much. Attempting to further drop the output by reducing the power supply voltage worked--to a point. The output voltage dropped a little but then went to zero, probably because the PWM chip quit working. This is problematic because the designs I have seen on the web just use their inverter's Vcc input as the "control voltage" in a feedback loop.

Looking at the transformer connections I saw what appeared to be a type of differential drive, with the center tap of the primary connected to the input Vcc and the "outside" pins connected to a switch IC that would alternatively connect the transformer pins to ground . This is similar to what's called the "Royer" oscillator, see here . I thought that it should be possible to break the center tap's connection to Vcc and drive the center tap with a variable voltage to vary the output voltage over a much wider range. The rest of the inverter circuitry would be "happy" because it's getting the voltage it was designed for.

So I cut the trace connecting the center tap to Vcc and soldered a wire to to the trace that will be used to provide an external "gain control" for the inverter. I also thought it would be a good idea to add a bypass capacitor from the center tap to the local ground, to handle the inevitable inductive spikes that arise during switching.

This approach looks like it will work OK, with an interesting side effect. I noticed that the HVAC coming out of the inverter isn't nearly as "peaky", even when I connect the transformer's center tap to the inverter's Vcc (i.e. it's running full-bore). There must be some pretty hefty transients present on an un-modified inverter! I'm using a .47uF 200V ceramic capacitor to bypass the center tap and so far, so good.

While other flavors of CCFL inverters may use somewhat different circuits to do their job, if your inverter uses a transformer with a center-tapped primary you probably can use a similar approach to get better control over the HV output compared to simply varying the inverter's supply voltage. I had initially thought I would have to de-solder the transformer pins and lift them off the board but then I found a circuit trace I could cut to accomplish the same result.

Note that the classic Royer Oscillator uses a saturable transformer core, so dropping the voltage on the center tap _might_ be problematic once the current is too low to saturate the core. Inverters that use a PWM chip shouldn't care too much about that.

 

[graham-xrf]

Installing Kicad 7 AGAIN!
It seems to take way longer than I thought to get everything back to the way I liked it. Fortunately, all the design stuff so far was all there, unharmed, in the recovery backup. I just have to fix things again so that the symbols I made are in a library I can access and edit. As best I can, I am pressing to get this kit to the point it experiences at least some electrons moving, even if it blows up!

I really do want to discover whether we can muster enough high resolution gain to make our idea competitive with a photomultiplier tube. I think what might have to be appreciated is that when one uses low pass filtered pulses, the spectral content will have lost information. Preserving the pulse waveform as much as possible might result in valid integrated areas (under the curve representing the photon energy) as being slightly different to another pulse that looks much the same if both were filtered.

The problem is, how much to smooth it out, to throw away meaningless high frequency noise racket, while not taking it to the point the integration gets sullied. I begin to think the high resolution discrimination is about being able to reliably count up the value of the energy in a pulse, and know that it can be different to another very similar looking pulse, representing a different element in there.

 

[homebrewed]

Speaking of a "racket", the CCFL inverter I'm trying to use as the basis for a PMT power supply certainly makes one. It's going to be challenging to get a really clean voltage. I think most of the noise is coming in on the control line -- basically, the center tap of the transformer primary turns. Before I put a home-made choke in series, it was bad enough to destabilize the feedback loop. I made a "barrel" type choke to filter out the noise -- I turned a piece of HRS down to 10mm dia x 10mm long, drilled/tapped a center hole and attached some 1"x1" CRS end pieces. Nothing like mixing measurement systems, huh. I chucked it up in my lathe to put the windings on the choke. The metric dimensions make it easier to calculate the approximate number of turns, maybe something around 100 turns. There still is some noise getting through, probably due to turn-to-turn capacitive coupling and the fact that the core is a relatively crummy solid hunk of low carbon steel; but now the feedback loop is working and I can adjust the HV up to at least 1KV. It's possible I'm getting some magnetic coupling from the transformer so a piece of thick-enough aluminum (>> skin depth) might help, too.

I'm thinking about a different approach, using an optocoupler to isolate the CCFL inverter even more from the control circuit. A dual Darlington would be ideal -- one would be used to control the CCFL and the other would be used in a feedback loop to linearize the optocoupler's transfer curve. The LEDs would be series-connected so they would see the same current.

 

[homebrewed]

It turns out that a lot of the residual noise I was seeing on my oscilloscope was switching noise from my work bench fluorescent lamp! It probably has an inverter in it, too....

 

[WobblyHand]

It turns out that a lot of the residual noise I was seeing on my oscilloscope was switching noise from my work bench fluorescent lamp! It probably has an inverter in it, too....

I've been fooled by that before, too! Wall warts, led lights, just about everything has an inverter in them. Some are horribly noisy, and radiate a lot. Found that when I was playing around with my doppler chronograph. Cheapo wall wart supplies, had to unplug them one by one, until the really bad one was found. I kept on getting this weird 12-13 KHz signal coming through. Somehow the signal was coming though the ground or neutral line and coupling into the DC lines. When I unplugged that unit, all the noise vanished.

 

[homebrewed]

It also looks like the TFT display on my DSO is radiating noise, too. I'm trying to measure "real" PSU ripple & noise down in the 10mV range and at that level there seem to be all manner of things radiating noise. We've got lots of LED light bulbs in the house, each with its own little switching regulator injecting noise into the AC lines. I may get up in the night and see what the noise level looks like with most of them turned off!

 

[graham-xrf]

It also looks like the TFT display on my DSO is radiating noise, too. I'm trying to measure "real" PSU ripple & noise down in the 10mV range and at that level there seem to be all manner of things radiating noise. We've got lots of LED light bulbs in the house, each with its own little switching regulator injecting noise into the AC lines. I may get up in the night and see what the noise level looks like with most of them turned off!

Very likely, the route in is by conducted interference carried in on the 0V end of the powering arrangements. If the rest of the kit is reasonably electric field screened, then there remains the possibility of low frequency ingress by magnetic coupling, but that would be 60Hz from house wiring and appliance cables. Switcher racket from LED lamp mains step-down electronics has higher frequency blips, and finds it very easy to contaminate the house mains, the wires being quite effective transmission lines. I guess it might be time to bring out the common mode chokes and suppression measures.

It is actually possible to prevent such noise riding in on the conductors. Provided the kit has reasonable electric field screening, you can expect to be able to eliminate the switcher waveform contamination, and let the lamps keep making a racket.

Even getting a scope to confirm the noise on the trace is real can be a challenge. I have found I often cannot just clip the ground of the probe to a 0V, and the common way of allowing the circuit 0V to find it's way onto frame metal would likely defeat attempts to get the signal clean. In my stuff, the usual effect of such grounding does, to a reasonable degree, reduce the coupled trace noise wobbles to a certain low(ish) level), which then resists attempts to clean it further until the circuity is isolated from the frame. Sometimes you can see "PSU switcher noise", and discover it is just a measurement artifact that goes away when dual differential probes are used.

TFT display, being so intimately connected to measurement circuits, might perhaps need to be separately cleaned up. I think I would be unhappy with as much as as 10mV, but I do accept that the 500uV to 1mV level is a darn hard thing to achieve.