Author Options:

I'm having unknown issues getting an ultrasonic cleaner hack to function properly - send an electrical engineer. Answered

I'm hacking a cheap ultrasonic cleaner to run for an indefinite period of time instead of 8 minutes max. It works well enough with a switch soldered directly onto the main board, but I intend to leave it for hours and I can't have the thing burning out so I've got an attiny85 running as a simple timer (can't get a 555 running accurately for the periods I need). The ultrasonic transducer is controlled through a transistor and relay which I'm using to my advantage.

I've got it hooked up to the built in 5v supply through a switch on the ground and flip the attiny pin tied to the transistor high. Really it's all tied  right to the brain board connector for convenience. It only sort of works. I've tested the attiny to death so I know the code works, it's just the arduino blink program with numbers changed.

When I flip the switch, it supplies power to the attiny and I hear a transformer buzz and the xducer fails to turn on. The screen doesn't dim any which I'd expect from being underpowered. When I hit the machine's on button it does turn on and when I hit the off button or let the time run out the machine stays on (small success). Once the timer cycles though it won't turn back on, which functionally makes it like I've added a 25 minute timer.

I suspect it may be something to do with not enough current to get the transistor started, but enough to keep it conducting. Perhaps if I connect the attiny directly to the base of the transistor instead of through the 10K resistor on the main board it'll work properly. Then again I know basically nothing about what I'm doing and should feel lucky that I managed to hack it this far without electrocuting myself.

Update: It get's weirder. I stuck a 20k pot on the output of the attiny and hooked it directly onto the base of the transistor, bypassing a 10k resistor, and it runs constantly regardless of the inputs. I hook it up before the resistor and the behavior remains as before.



2 years ago

Hello Sirs,

I have a problem with the board of the unit. Moisture made one resistor to burn and I cannot see the color code on it. I cannot see the resistor on the picture of the board posted before because is to near to C4. The resistor is R3 and it is the bottom-right part of the board as is captured in the picture. Can one of you be so kind to read the color code from the board and to post it here. I can identify RED-RED-UNKNOWN-GOLD. The third is the most important. Thanks a lot!

Here's a partial schematic because the board it's hooked to is weird
and hard to trace. The only chip on the board is a CD4820 and I can't
find anything on it. My bit is literally just a switch and an attiny on
perf board with massive ugly globs of solder that I managed to keep from
bridging. I'll try to draw up a more complete schematic but all I can
do is probe with a harbor freight multimeter so no guarantees.

don't want to do anything that I can't remove later which is why I've
not gone with a thermometer controlled system (I tried because it's got a
built in thermistor), but I would be willing to build an entirely
different control board (if I can find the connector).


Yes. I did include a photo that pretty clearly shows it. More reasons: the power pins on the chip in question are in the middle of the chip instead of on opposing corners and tons of nearly identical ultrasonics show up for sale when you google CD4820.

Middle IC power pins suggest a quad op-amp if it is a 14 pin chip.

It can't be though, it's driving the display with most of the pins.

I think you might be on to something with that. I imagine it's just a rebranded/counterfeit one of those since a production run of custom chips would be too expensive for a <$100 product (mine was $60). It would explain the lack of a proper datasheet too.

I should have said rebranded the Chinese are notorious for that, resin coated ICs are custom.

Yes you said and I quote, "The only chip on the board is a CD4820 and I can't find anything on it."

A lot of people call ICs chips.

I don't see how a terminology discussion is helping me to get the bloody thing working when you understand what I mean either way. I think I'll add "magic pixie wrangler" to the thesaurus.

The first time I looked at the photo it did not show the IC so I was wondering what you were talking about.

I am blind so I blow things up in order to see them, the chip was off screen when I first looked at your picture.

Here's a mostly correct schematic, my apologies if it looks terrible. The bit in the blue box is what I'm having trouble making work and all I want to do is make that transistor in the corner turn on and off.

I screwed around some more and found that if i disconnect the ground from the control board, leaving the 5v and atmega pins connected, the display flashes on and off.


I tried that as well and it didn't work. I suspect when that my first R1B value of 0 ohms killed the chip. I also tried adding a pot so I could find an R1B value that worked well enough and a 100uF cap across the power leads to fill in the power spike but I broke it before I could make that work.

I think I busted a second one on the breadboard because it won't talk to the programmer correctly.

Does the circuit board you made the connections ions look something like this, (Doesn't need to be exact.)


OK now I get it, nice picks I can even read the writing on them.

CD4820 is the model number it is written on everything, the circuit boards and the custom IC on the display board. That is why you cant get a datasheet on the IC, it is a custom IC. I hop you can see the red boxes.

Most likely it is a onetime programmable IC.

The only thing I can suggest live reverse engineering it



2 years ago

Triacs once in conduction stay in conduction except for four cases five if the weather be good..

Case # 1 ] AC phase control when the AC voltage reverses against the conduction direction it the BT137 triac will turn off = block conduction..

Case # 2 ] DC Holding current threshold, if a BT137G latched current falls below 40ma the triac will self turn off and block conduction...

Case # 3 ] Communicating condition by another semiconductor which draws the current through a different and lower impedance path, long enough to turn off and block the triac conduction....

Case # 4 ] Extremely Rare Gate turn off device only seen in the SCR by this eng.....

Case # 5 ] I^2xt fusing current burns the device open never to conduct again.


GTO is an SCR silicon_controlled_rectifier... Long time ago you could fire an SCR turn it on into conduction with a narrow +5ma for 20us pulse gate-cathode...The SCR will remain in conduction with no more action on the gate and the kathode = cathode....

Then if you want to turn the SCR Off just pull -250ma from the gate for about 10ms and the GTO will stop conducting...

The reverse gain is 50 less was a no go....

Good now what do the hall effect sensors in the motor do?

They take over the equivalent of the brush and commutator of a DC series universal motor by switching alternate stator windings by detecting a rotor magnet and energizing a coil to pull or push the magnet.

On bidirectional coil windings a dead moment needs to allow transistors to turn-off before turning-on the opposing transistors.

Hall PM.pngPM FAN.jpg

IIRC they were a pain to use in practice, and were not that common. There are well established methods to commutate SCRs, because I remember spending an interminable amount of time in a power electronics class in engineering school learning about them.

These days, high power/high voltage mosfets, or insulated gate bipolar transistors have come to the fore, but the thyristor still wins at very high power levels - grid level inverters are still thyristor based. I remember seeing presentations on the motor drives for the TGV in France, and they had megawatt devices then.

BTW looking at your triac circuit running a brushless fan on DC.

I would wager the brushless fan is switching to maintain prop speed and during the low switch current draw slips below the holding current long enough to turn the triac off !... when you are not actively firing the opto-isolator to the triac gate.

Funny I never received a notice of this comment.

Yea the Triac is being turned on and off with the hall effect sensors in the motor.

And the circuit works.

I could be argued the motor turns its self off.

Funny I never received a notice of this comment.

Yea the Triac is being turned on and off with the hall effect sensors in the motor.

And the circuit works.

I could be argued the motor turns its self off.

I've got a few 50A SCRs in a bin somewhere and I investigated how to use them and similar devices thouroughly when I was building a coilgun for school because I didn't want to spend the $20+ on IGBTs or other parts (I used some supercaps to test whether they did better than electrolytics - they don't).

AC hardware is not going to work in the place I'm working on.

Correct the Communicating it should say Commutating...

A CSR or a Triac might be better than a transistor.

A transistor used as a switch must be run at saturation but no resistor can burn out the ATtiny.

What is the transistor?

Doubtful. The transistor is what's built in and since it's not controlling anything AC a triac wouldn't turn off.

That is funny triac's turn things on and off for me.

That aside picks and schematics so we can visualize what you are trying to describe in stead of guessing helps.

What ever you do do not power the transistor base without a resistor or it will fry the ATtiny I wouldn't go less than a 100 ohm.

FYI Triacs work on AC, and turn off every half cycle. Used on DC, a thyristor, which is "half a triac" won't turn off ever, until there is no volts across it, or no current.

That's because the triac is so poorly turned on, it isn't turned on. Look it up. You aren't injecting charge into the gate, you are allowing charge to leak from the gate.

Please don't tell people you can turn off a triac when its on, you can't except in the ways I indicated.


TRIACs are a subset of thyristors and are related to silicon controlled rectifiers
(SCRs). However, unlike SCRs, which are unidirectional devices and only
conduct current in one direction, TRIACs are bidirectional and conduct
current in both directions. Another difference is that SCRs can only be
triggered by a positive current at their gate, but, in general, TRIACs
can be triggered by either a positive or negative current at their gate,
although some special types cannot be triggered by one of the
combinations. To create a triggering current for an SCR a positive
voltage has to be applied to the gate but for a TRIAC either a positive
or negative voltage can be applied to the gate. In all three cases the
voltage and current are with respect to MT1. Once triggered, SCRs and
thyristors continue to conduct, even if the gate current ceases, until
the main current drops below a certain level called the holding current.

Sorry the power to run the motor is not coming out the gate, if it was it would fry the photocoupler in a nano second from over current.

"the triac is so poorly turned on, it isn't turned on." I mean really?

Eh ? The gate current doesn't drive the load, the gate current is just like the base current in a BJT.

And yes, really, look it up. The wikipedia piece is excellent.

Just don't tell people TRIACS are switchable on and off, they are not.

I read it and wilipedia calls a triac a switch even the datasheet say switching.

Wiki or Philips Triac PDF .... NEVER .... talk about switch-off !..

Wiki references Switchxxx-on six times and the Triac PDF does reference

switch to the on-state.

It switches off also, I built it and it does it, and it is not powering the motor through the gate because that would be 800% the power capacity of the photocoupler, that is why I used the triac in the first place.

No, it CAN'T be powered BY the gate, its powered by current flowing Mt2 - mt1, but either the gate current is too low to turn it on, your gate only leaks through the photo transistor, or there isn't enough current in MT1-MT2 to latch the triac on properly.

What do the hall effect sensors in the motor do to the Triac Steve.

The Triac is getting 12 volts 400 mA and then it is getting nothing, then it is getting 12 volts 400 mA and then it is getting nothing. So as long as the gate is triggered the Triac fires with the hall effect sensors in the motor.

Stop triggering the gate and the Triac remains in an off state the next time the hall effect sensor in the motor turns off the current.

The circuit works.

I believe you..

If you still have the circuit, you could substitute an incandescent lamp for the fan and shoot my idea of pwm holding current drop out of CANADA :-)

I've got a partial schematic and I'm not sure how much a full one would help since the cd4820 chip that does all the magic is a mystery to me and digikey.