Test Ultrasonic Inverter with No Load

I don't see a problem with this, the inverter will typically draw about 1-2% of their rated capacity at idle(no load)...just be careful not to short the two leads together, not sure what would happen in that case....diodes do wear out

"Over time, a diode that has degraded tends to exhibit an increase in the reverse leakage current while the device is operated within its rated voltage range. Related changes in the diode's specifications could include reduction in the device's breakdown voltage and forward voltage drop."

• If diodes are subject to current overloads, their junctions can be damaged or destroyed.

https://www.mdpi.com/2076-3417/13/12/6991

Another item to consider is the temperature bath of the cleaner, the resistance and current draw should be measured at operating temperature......A piezo material is basically an insulator. That means the internal Resistance is extremely high. To measure it we need a so-called Teraohm-meter. This instrument indicates the resistance in decades like 10⁶, 10⁷, 10⁸, 10⁹ Ohms etc....

..." The picture shows a few piezo-materials with their sensitivity versus the maximum allowed operation temperature.

It shows the drastic decrease of basic sensitivity with increasing temperature capability of the material.

​

Lead Zirconate Titanate (PZT) is the most commonly used piezo ceramic with compression mode charge constants of a few hundred pC/N and reasonable operations temperatures.

PNM-PT is a group of recently developed single crystals with ultra high sensitivity. They show charge constants of 1000 to 2000 pC/N but with Curie-temperatures of 30 to 80°C.

​

The Bismuth Titanates (BT) are ceramics in the 500 to 600°C range with sensitivities of 10 to 20 pC/N

​Finally there is a real high temperature group (all of them single crystals). Here we find Langatite and Langasite (LGT, LGS), Gallium Phosphate (GaPO), or the rare earth calcium oxoborates such as YCOB, with about 4 to 8 pC/N. This group includes also the natural crystal Tourmaline with 2 pC/N....

https://www.f3lix-tutorial.com/piezo-materials

https://ramkleen.com/industrial-parts-washer/?

Thanks. The cleaner uses a water-based cleaning fluid, which is heated, but no where near boiling.

You give no clue as to the power in watts, input voltage to inverter or output voltage/frequency. Voltage & fusing or transistor rating should give a clue.

I guess the frequency is ~ 10 kHz and the inverter has a transformer. I have used a sound audio amplifier to drive HF transfos to check the ratio. You can start at low voltage & 20W up to 20 kHz is quite common for Hi Fi amps. Will give you an idea of working voltage you can compare with component ratings .

Inverters usually have some zeners or VDRs to clip overvoltage spikes at transistor switch off - should be possible to identify such parts.

Peizo transducers will look like capacitors until the mechanical loading absorbs energy. The circuit might have precautions against blow-up if someone powers up with no fluid.

I only have access to the inverters, not to the rest of the device, so I don't know the precise power involved. On the other hand, I'm pretty sure that each unit is connected to a standard 120V supply, and it has a 5 Amp fuse, so the power consumed must be well under 500W. Switching DC power supplies of similar size have a power output around 50 Watts. A photo of the failed unit follows, with the known failed components removed.

The board is 3.5 x 4.4 inches. The output transistors are D13009s, which I gather (I haven't been able to find an actual spec sheet) are rated at 12 Amps and 400 Volts. The space for C21 is vacant on all three boards. The diodes D1 and D7, blown to shorts, are FR217s, rated 1kV @1.5 Amp. They are in parallel with the E-C of the output transistors. I've never seen a fuse blow like this:

Maybe lightning got into it....it can have some weird looking effects...

Hi DKWARNER,

Reply to your post #6.

This is my interpretation of yr picture of the PCB.

1/ This is a self-oscillating push-pull inverter.

2/ The yellow bits bottom left are a filter L & N to keep HF out of the line. The blue blob may be a VDR voltage dependent resistor to clamp line transients.

3/ DB1 is a full wave bridge rectifier - probably no more than 1 amp rating.

4/ C10/C11 are probably DC reservoir capacitors on bridge output.

5/ The burnt D7/D9 are zener diodes collector-emitter of the transistors [emitters are on supply -ve]. To protect against overvoltage spikes - it's odd you write they are 1 kV, but transistors are 400V - 1kV believable for 220V version, but this is 110V - cause of failure?? do the good inverters have different zener?

6/ The collectors connect to a ends of [centre tapped] primary of transfo [probably DG2]. Centre tap goes to DC +ve.

7/ DG2 may have a centre-tapped secondary with ends to the bases of the transistors to give positive feedback and oscillation as well as a separate output winding. The centre tap goes via a base current limiting resistor to DC negative & transistor emitters. Or maybe... see 8/

8/ This kind of inverter often had a separate saturating feed-back transfo with tapped secondary for transistor bases. Its primary was collector to collector of the transistors with a series resistor [R11?] to limit current when it saturated. This avoids magnetic saturation of the main transfo which can overcurrent the transistors since there is little winding resistance to limit the current. SY2 top right may stand for "saturating...." [note it is labelled 110V - 220volt would be a different part]. It also meant the frequency could be controlled by the saturating transfo avoiding magnetic saturation losses in main transfo.

9/ C8,R8 and C12,R? are probably series R-C snubbers to limit collector voltage when switching each cycle.

10/ The toroid near the middle may be a DC filter choke with C10/C11.

On the other hand SY2 might be an inductor to tune the capacitive ultrasonic transducer to inductive - capacitance loads are not kind to inverters.

Looking at the PCB tracks not visible in the picture should clarify.

Should say that you can probably load on bench, having no transducer, with 25W resistor load @ 220 volts collector-collector - voltage of secondary to ultrasonic transducer unknown, see post #4.

Thanks for taking the time to analyze it!

25W resistor of what resistance? 25W @220V would indicate about 2kΩ.

AFAIK, the unit runs on 120VAC; Since there are no large capacitors (C10 and C11 are 1µF, 450V) the peak DC voltage could be well above 120V. The diode bridge is rated at 2A, 1000PRV, and it too has failed to a direct short, separated from the input AC only by the ≈0.1Ω DC resistance of each side of the input inductor, labelled T2.

Since you hinted you want to see the circuit traces, here's the back of the single-

sided PCB: (Strange placement)

DG3 is indeed the output transformer.

Hi 67,

GA! Thanks for your effort! Your drawing does help. Since you got me started, I'm working on a complete circuit diagram, and will post it here, hopefully tomorrow.

It turns out that DG2 (Not DG3 as I said in the previous post) is in fact a single winding inductor (it IS labelled L2), not a transformer. The pin whose trace connects to the -output does not connect to the winding.

I just realized that SY2-110 has the -110. It is a two winding transformer with no extra terminals, so a different windings ratio would be used for the 220V version.

I suspect that the symbols you used for D7 and D14 represent Zener Diodes, but they are FR157 Diodes; 2A @1kV.

Reply to dkwarner.

Seems the transducers are capacitative. For driving, this is dealt with by a series inductor to tune out the capacitance.

Data for a 60 watt transducer for 25/40 kHz I looked at gives ~4000 pF and resonance resistance 10 - 20 ohm. For 25 kHz, 10 millihenry would resonate. Inductor reactance would be about 1.5 kilohm suggesting transducer and inductor get kilovolts across them.

https://www.bjultrasonic.com/shop/25khz-40khz-60w-double-ultrasonic-cleaning-transducer/.

Measuring the ratio of the transformer and inductance of L2 and getting a transducer data sheet/frequency from the user are necessary - the frequency should come from L2 and xducer capacitance, verified by user.

For a test load you would need a capacitance similar to xducer in series with a resistor similar to its upper resonant resistance initially, lower for soak test. When L2 and frequency are known, capacitance voltage/VA rating can be decided.

The state of the blown fuse is a concern - it does not seem it was capable of breaking the fault current from the line. I believe a typical 120V 60 Hz outlet would be 40 amps load rating. The breaking capacity of wire in glass fuses is only about 30 amps at 230 volts - data I have on UL wire end 2.3 x 8mm ceramic fuses gives only 50 amp @ 125V AC - can you tell what dimensions/type from undamaged boards?? Ceramic high break capacity HBC fuses are better - 1" x 1/4" 1 to 13 amp fuses used in most "domestic/similar outlet" plugs here are rated 6 kA prospective @ 250V.

Thanks again!

I have never seen the cleaner nor its transducers, but they tell me it is an inexpensive unit. Since it has three driver boards, I presume that it has three transducers. At 65mm and 500g, I doubt if the transducer of your link is inexpensive.

The Main unit is in Southern California, and I'm in Northern California with the driver assembly, so there is no way I can connect the presumably good drivers to their transducers.

Since the capacitance of the transducer presumably produces resonance with the inductor, and it would be that resonance that produces the high voltage, It sounds like the absence of the transducer would prevent resonance and therefore not produce the high voltage, so I'm leaning toward powering up one of the good drivers and putting an oscilloscope across the output terminals. I haven't done that yet.

The fuses are Ø4mm, 11mm long, with a 3mm gap between the two end caps.

I've been told that the cleaning solution is water with an unknown amount of a cleaner they call "Blue Gold", which is "mildly caustic". I see no evidence of liquid contamination of the blown board, but of course "mildly caustic" means the solution has some electrical conductivity, so contamination on the fuse could have initiated the arcing after contamination elsewhere initiated the failure.

I still have a bit more to do on the circuit diagram...

No output current = no current in toroid primary to oscillate - oscilloscope no good.

By my estimate the transfo driven by transistors is actually stepdown to 10s of volts - xducer 20 ohms resonant for 60W, say 1.5 amps current, 45 watts & 30 volts. Check the ratio if you can.

You might try 30 ohms resistance [3 x 10 ohm 10W] on secondary of SY before L2 [make sure toroid winding in series] - the inductor L2 may take too little current.

The xducer I referenced is $12.

67model

"No output current = no current in toroid primary to oscillate - oscilloscope no good."

Oh, that's right!

More after I finish the Circuit diagram...

OK, I think I've got it! Here's the circuit diagram I've made:

I've measured the DC resistances of all the inductors, having subtracted the lead resistance to obtain all values shown.

I think you are right: Without the capacitance of the transducer, it probably won't oscillate at all, and certainly not at the correct frequency, so applying power without the transducer would be totally meaningless.

Once again, thanks for your expertise!

If the User does not tell you the transducer make/type/frequency, it is difficult to test.

However, if you measure L2 milliHenries & assume 40 kHz you can calculate the capacity to resonate.

Taking the 3900pf of the transducer I identified, 3 kV capacitors are not common & cheap ones cannot carry 2 amps.

Brutally, 40 x 40 cm aluminium foil plate pressed onto bare copper wire on two edges is ~ 7 sq.mm & well able to carry amps. Two plates separated by 1 mm of PET [used for transparent plastic food bags & trays] would be ~4000 pF. Weigh down with plastic box/rocks. Theoretically 17 kV/mm but layers & air will reduce. I measured 4 heavy duty freezer bags [8 thicknesses] as 0.5 mm so could try with 30 x 30 cm.

dkwarner,

vsar's emphasis on water loads has triggered a thought!

The permittivity of water is about 80 so 60 sq.mm "tin can" plates in water with 1mm separation would be about 4000 pF ; some dimpled plastic for 1 mm separation. Adjustable be sliding plates apart and water cooled! Pure water is a good insulator so start with tap water, but I have not found dielectric loss yet [perhaps vsar can help].

You need to measure L2 inductance to estimate the capacitance for 40 kHz? most common cleaning frequency.

67model

It is low till UHF and dielectric constant ~87!

And as vsar writes you can adjust by raising plates out of water!

The dielectric loss of water, I think will become apparent as current passes through the water due to the temperature changes in the medium...

During the test, the medium concentration will change which is directly proportional to the length of time and amount of current injected into the solution... Ultimately until the solution starts or reaches the boiling point, akin to how microwave heats the food.

Sorry it's more than 50+ years now and can no longer recall any computational details related to the crude test load that I rigged together..

vsar,

Thanks for the reply, I did find water data [post #28]. My memory of ∈ = 80 was correct @ 20'C! Interesting it varies a lot with temperature.

I forgot to acknowledge the source, it was....

https://water.lsbu.ac.uk/water/microwave_water.html

Water behaviour is very complex!

Some thoughts for dkwarner.

If you measure L2 inductance & assume commonest peizo frequency of 40 kHz you can add series resonance capacitance [see post #27 for water capacitor] and load resistance ? 20 ohm? as transducer substitute for a test.

How to measure inductance of series inductor L2...

1. ground junction between output transfo SY and inductor L2.
2. Connect common clip of oscilloscope lead to junction of 1.
3. Connect input of oscilloscope to OUT + terminal - the oscilloscope is reading the voltage across inductor L2.
4. connect common of signal generator to common point of 1.
5. connect signal of signal generator to OUT+ terminal via series 470 kilohm resistor.
6. Connect 1000 pF capacitor C across L2. Close tolerance +/- 1% polystyrene or mica is best, the common Mylar 1 nF is +/- 10%.
7. Set generator output to a volt or two.

You now have L2 and C as a parallel resonant circuit, the 470k resistor ensures the generator does not damp the resonance [the scope input resistance usually 1 megohm adds some damping]. As you vary the frequency, you should get a distinct response peak on the oscilloscope at some kHz - about 50 kHz for 10 mH & 1000pF.

10 mH has a reactance X of about 3 kilohms at 50 kHz, so resonant impedance X times Q would be 3 x 10 = 30 kilohms for a modest magnification factor Q of 10. Note that 470k is much greater than 30k so it will not damp out the resonance peak [but you will get about 50 mV peak on oscilloscope for 1 volt from generator]. It is likely Q >> 10. Also note that Q is peak frequency/bandwidth at 0.7 x peak response, so if period changes 10% between the two 70% responses then Q is about 10. Easier if the generator has accurate scale or frequency readout.

You can get the frequency f from period P of 1 cycle on the oscilloscope, f = 1/P e.g. 50 kilocycles would have a period of 1/50000 seconds = 20 microseconds

The resonance frequency f₁ is 1/[2π√(LC)] at which inductive reactance = capacitative reactance ( note the π is 3.142, a not nice pi symbol in CR4). Since C is known and f₁ is measured, L can be calculated.

L = 1/[39.48 x f₁² x C] N.B. L Henries, f₁ Hz, C Farads

Or less "calculator bending" L henry = 10⁶/[39.48 x f₁² x C] with f₁ kHz, C pF

With f₁ 50 kHz, C 1000pF, L is 0.0101 Henry (say 10 milliHenry)

You may have noticed that L2 must have some self capacity C_c , so C is really 1000pF + C_c [including scope capacity - easier if scope on x10 ]_. But you can take off the 1000 pF and find the new resonant frequency f₂.

f₂ will be higher than f₁ so:-

f₂²/f₁² = [1000 + C_c]/C_c from which C_c x [ (f₂²/f₁² ) -1] = 1000 pF hence C_c in pF.

C_c will be inductor capacitance plus scope & cable capacitance [the 470k resistor is about 1 pF & can be ignored]

1000 pF can be replaced by whatever known value put in parallel with the self capacity C_c of L.

So you can estimate the inductance, its self capacity and quality factor Q from the measurements.

WOW! Thanks for the info, both from Vsar and from 67Model!

I do have an old but very good CRT Oscilloscope (Tektronix 2236), a function generator good up to 100 kHz, several multimeters, and a pretty large collection of components, so I could probably do this, but I'm not convinced that it's worth the trouble.

I have received most of the repair components, but am waiting for a couple more. I'm inclined to simply install the damaged components, then compare many resistances between the repaired board and two others that are presumably good.

"I do have an old but very good CRT Oscilloscope (Tektronix 2236)"

Get your company to buy you a DSO.

Or treat yourself to one:

You should be able to buy a four channel 100 MHz analogue bandwidth "hobby scope" for about $250.

Something like this https://www.youtube.com/watch?v=-Ov1cddCGO4

Wow! I had no idea that such a unit had become so inexpensive. Thanks!

On the other hand, I am old enough that I am only doing occasional design and testing, and my faithful 2236 has served my needs every time I've used it, and its built-in multimeter is still the best one I have. It's what I used to measure the resistances to a precision of ± a few hundredths of an Ohm.

Yes, if client will not pay for test, not worth trouble.

Fuse not satisfactory - obviously short cleared by line breaker not PCB fuse. Investigate with client line breaker size/type & reduce if possible. Wire-in fuses seem to be USA/UL standard of limited break capacity. Covers your back! Far more people die of electrically-started fires than electrocution. It burns down businesses too.

I agree! The replacements I got have a 10mm glass gap between the end caps, 2.5 X the original gaps.

As long as you know the rated power capacity of the inverter/power supply, you can always rig a corresponding matching load... using brine solution..

I remember connecting my inverter output onto steel rods soaked into the brine solution.. You can vary the current being drawn by the solution by varying the immersed depth of the rods..

In my testing load, the rods were mounted on a plate made of plastic material which can be cranked up /down into a plastic tank of salted water solution..

In my opinion, the power supply needs to be tested to the maximum designed power handling capacity, impedance and duration..

Good luck..

Unfortunately, I do not know the rated output, and I don't know the characteristics of the (presumably piezoelectric) load, except that its DC resistance is higher than a standard multimeter is capable of measuring.

I have seen carbon rods with variable immersion in brine used as a speed control for a large motor, but I doubt that would be a suitable substitute for a piezoelectric transducer of unknown size and characteristics.

Since piezo crystals behavior changes depending on the kind of power applied it receives, then you cannot just rely on its unknown DC resistance characteristics.

Being a dynamic load, the crystal will also generate harmonics that your power supply should be able to handle and tolerate..

This is the reason why I believe a variable brine solution test approach will be applicable, as long as you don't exceed the rated amperage of the power source..

Depending on the crystal's generated harmonics the current drawn to the solution during the test can be expected to fluctuate especially when the solution starts to heat up and boil..

That actually is the beauty of using the brine solution as the test load, beside being dynamic, all you need to do is measure and monitor the current being drawn from the power source as you adjust the length of the submerged rods in the solution.
You can also add a current limit through the concentration of salt and water.

I agree, except that I have no idea of where to start. How much salt in what size container with how much water, and what size electrodes immersed how deep for an initial value?

Any 5 gallon plastic bin half filled with water and salt mixed together should suffice your testing needs..

The inverter output separately connected to solid metal rods like #10 copper wires, that are mounted on a plastic plate... The wire / rods long enough to enable being dipped or lowered into the bin of brine solution..

The test setup would require a preconnected ammeter and volt meter at the inverter output as you slowly lower the plastic plate mounted rods into brine solution..while at the same time observing the current meter readings..

It would be apparent that both the length of the rods submerged and time together will reflect the amount current drawn from the inverter..

You may need to add more salt if no current flows.. Or if the brine solution starts to heat up and boil immediately then the solution is too salty and more water is needed..
This test should only last for a couple minutes, unless you want to test and validate the integrity of the repairs made..

Thanks. I have all those items, but since the circuit apparently uses resonance to generate the high voltages that drive the piezo transducer, I suspect it won't work without the proper capacitance.

Take a look at the circuit diagram I just sent in reply to 67Model's recent post.

Agree since the capacitive reactance of the solution is dependent upon the salt concentration. Which means that you have to start with a very low amount of salt to get a very high capacitance between the electrodes using a very low salt concentration as the dielectric.

That looks like a very powerful surge to me.Lightning or wrong voltage applied.

It arced across the top of the glass fuse,so it had to be a very high voltage and current.

All of the components are suspect, even if they check good now they could fail prematurely later.