Ultrasonic driver discussion. 1

[itsmoked]

Remember my recent "help me find an ancient transistor" thread;
thread240-443861

I'd like to understand these boards a little more, please walk with me as I describe the scene.

Here's a reminder of what they look like:

I referbed five of these things and laying on my stomach installed them into their 19" rack-mount boxes stuffed under heated cauldrons of 409-like detergent and DI water. Bolted to the bottom of these 2ft x 2-1/2ft x 2ftdeep tanks are the four ultrasonic transducers under a heavy SSteel cover.

Each of these 19 inch racks had 230V 80mm metal computer-style fans. One was failed in one of the racks probably leading to the worst vaporized caps seen on some of the boards.

One I'd refurbed they'd installed and had re-failed, in the last week before I got there yesterday. I was surprised to find they were 230V not the expected 120V and it's 10A fuse had blown, a bad sign. :/

I pulled out the board and checked the base-emitter diode drop across one of the 6 paralleled transistors and it was measured as 0.01V in either direction and both the B-E and B-C junctions. I decided it was a shorted transistor which will light up a fuse in a big hurry. I unplugged the transistors and yep, correct. One had turned into a three terminal 0.1Ω resistor.

I installed the four boards I'd brought and everyone of them fired up and made the cleaning table sound like it had 30 million angry bees in it.

Interestingly they have 4 red LEDs on the rack front panels that tell you each channel is driving its transducer.

They used those LED driving current coils and ran a lead to each transducer thru each coil. I liked its simplicity and galvanic isolation. I left them all run for an hour while I troubleshot a problematic VFD that was, luckily, 10 feet away instead 400 feet as usual.

Once they all seemed to stay running I disconnected the one without a working fan until they get the replacement fan installed.

While I was there I decided to scope out what was going out to the transducers. It was hard. Not knowing what to expect I swapped to my TEK high voltage probes (1000V) that I got optional with the scope a bazillion years ago. I had to hook the ground to a wire I stuffed in the back of the Molex plug and then I had to hold the probe tip against the neighboring pin on the back of the board without touching or slipping while doing a no-handed push-up and poke the freeze button on my scope. 'Don't talk to me and go away guys!'

Here's what I saw..

~32kHz driving the transducers at 1,700 Volts pk-pk! I then noticed my scope probe would draw an arc as it got close to the pin. Sheesh.

If I reduced the sweep I got fish..

Further sweep reduction - schools of fish:

These patterns looked the same on my boards as it did in these pictures of a yet-to-fail untouched factory board.
Anybody know what's going on here? Why the envelopes? Deliberate or incidental?

I hope the 1,700V isn't across the 1,500V transistors.. But with a fist full of transformers on the board (see above) I'm hoping not.


And on a different bent do you have any reasons I can't make a 32kHz oscillator, (555? or divided crystal), amplify it with a BIG MOSFET and drive the 1.00Ω transducers thru an impedance matching transformer, avoiding all the chokes and limited lifetime caps and swarms of transistors? The schools of fish make me balk/wonder.





Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

Bees are not so bad. I would be more concerned if it was wasps. They are man-slaughters, you know.

The envelope is what you get when undersampling. 50 kSa/s will only produce like 1.5 Sa/cycle of the 32 kHz signal. It is a bit below my fellow Swede Nyquist's limit. And he used a brick-wall filter at that. Less than ten Sa/period will not work well if you run without any LP filter.

Add visual alias (low resolution scope screen) to that. And your recordings are just what to expect.

Why not run a 555? All those square wave oscillators need to be tuned to the load's resonant frequency. And that doesn't work if you sett frequency externally. Things (like resonant frequency) change with time and, especially, load. So there must be a feed-back from the load circuitry. That is what the low power winding is for. Without it, you will be out of resonance and even if it doesn't kill - it won't clean, either.

Sorry to be so blunt with you. I think that you need it this time. And I know you well enough to think that you appreciate it.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

On a 60Hz supply the modulation envelope looks pretty close the period of a half-cycle - 6 cycles / 50ms. By any chance does this thing run from rectified AC with little or no smoothing on the DC bus?

 

[itsmoked]

I think you're spot on Gunnar on the aliasing. That scope has gotten me a few times on that. I didn't consider the screen resolution as part of the issue either.. Thanks!

I thought of the resonance issue too. Why not generate the oscillation with a micro and read back the power transfer with the same micro adjusting the frequency to peak the power transfer to resonance with a digital closed loop?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[itsmoked]

Scotty; It does indeed. Power comes in directly to a bridge rectumfryer with no obvious filter capacitor at all. The big fish is 120Hz on the money..

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

I've always attributed that sort of thing to aliasing

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[Compositepro]

The waveform looks very similar to the power supply of a halogen lamp I have, which converted 120vac to 12vac. It was actually 20 kHz with an envelope of 12vac. For the application there just isn't any need to filter out the 20kHz. In your case the 32kHz is what you need operate the ultrasonic transducers but there is no need that it be of constant amplitude. Thus the filtering caps can be eliminated, saving cost and increasing reliability. And, yes the crystal of the ultrasonic transducer would be part of a resonant circuit.

 

[ScottyUK]

I wouldn't be too quick to condemn the instrument - I have seen that waveform before on devices which had un-smoothed rectified power rails. The first time I blamed the DSO (a THS3024) until I saw it on my analogue scope too (a Tek 2465B). It looks weird because we tend to expect a DC rail have at least some smoothing, but not all applications need it. Induction hobs are another example of this type of circuit - HF carrier frequency, but modulated at power frequency due to the raw rectified supply.

 

[Skogsgurra]

Nothing wrong with the scope. It does what it shall. But undersampling is a fact of signal theory, usually called DSP today, and if you do - this is what you get.

Rectumfryer, that's a good one, Keith. We have that in our cars. But the salespeople call it seat heating. We call it rövbrännare, that is very close to rectumfryer. Especially if the thermostat runs amok.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[itsmoked]

I think in this case Scotty is probably right about what we're seeing. Those fish are 8.3mS long which is exactly the power half cycle around here. That, coming out of a bridge makes total sense. The transducers get to see the AM modulation which will reduce the average power they see and may actually help the cleaning with varying amplitude.

About the transducers, never seen them in this case but every single one of them measure 1.00Ω I expect them to be piezoelectric and to measure ∞Ω so what's up with that?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ScottyUK]

Gunnar - that made me laugh!

 

[ScottyUK]

Keith, Gunnar -

ST have an application note for an induction cooker which uses a resonant half-bridge running from the unsmoothed DC link. Page 24 shows a very similar waveform to that observed by Keith.