# Vibratory pump speed control

#### modi72

Joined Apr 23, 2023
9
I have an espresso machine with a 41 W 120V 60Hz ulka Vibratory pump. I 'm hoping to add a speed control to control water flow. From what I read, the pump has a diode inside. So with a frequency of 30Hz, it actuates a piston, pushing a bit of water out. A spring makes the piston return to its original position, ready for the next pump.

Some people use a triac in line with the 120V line to the pump to reduce the flow speed, but it has a limited range of speeds, causes weird noises and essentially only works by inducing a certain rate of slippage (missed pump cycles, so a bit erratic).

Another way to tackle the problem (I think) would be to purchase a single phase VFD, but it's a bit expensive and large in size (can't hide it in the machine). So I thought I could design my own single phase VFD. But while considering that, it occurred to me there may be another simpler way, but I'm not sure. So I'm hoping for your help/input.

I was thinking of a circuit that would detect the AC waveform, and depending on a input selector position, decide on how many waveforms to let through (using a zero crossing SCR) . So on selector pos 1, It would let through 1 of 10 AC waveforms. Pos 2 the first 2 AC waveforms go through the pump, but not the next 8 and so on. So instead of having a continuous sine wave that varies in frequency, I would have n x 1/30 second sine wave pulses with (10-n) skipped 1/30 sec waves.

Obviously , this would not work for a rotary motor, but for this type of pump it might. Is this a bad idea? or perhaps it already exist (and if it does please let me, because I haven't found it online). Thank you.

#### Ian0

Joined Aug 7, 2020
8,409
You don't get 30Hz by rectifying with a single diode. You get the same frequency as your mains.
You can run those pumps from pulsed DC, just by producing a rectified and smoothed DC supply, then switching the pump with a MOSFET. You need a clamp circuit like a flyback converter (using a diode and a transorb) so that the magnetic circuit can reset quickly.
You can then vary both the pulse width and the repetition rate.
(I have tried this with a 12V pump, in a special effects fog machine because they use exactly the same pump)

#### modi72

Joined Apr 23, 2023
9
You don't get 30Hz by rectifying with a single diode. You get the same frequency as your mains.
You can run those pumps from pulsed DC, just by producing a rectified and smoothed DC supply, then switching the pump with a MOSFET. You need a clamp circuit like a flyback converter (using a diode and a transorb) so that the magnetic circuit can reset quickly.
You can then vary both the pulse width and the repetition rate.
(I have tried this with a 12V pump, in a special effects fog machine because they use exactly the same pump)
Thank you. That was a stupid mistake about 30Hz, not sure what I was thinking.
I did not realize a pulsed DC would work too. Do I aim for 120V DC (the pump runs on 120V AC normally)? I imagine a pulse length of 1/60s is ideal. or does it not matter (I don't want to overheat the pump).
Thank you again for your help.

#### Ian0

Joined Aug 7, 2020
8,409
Thank you. That was a stupid mistake about 30Hz, not sure what I was thinking.
I did not realize a pulsed DC would work too. Do I aim for 120V DC (the pump runs on 120V AC normally)? I imagine a pulse length of 1/60s is ideal. or does it not matter (I don't want to overheat the pump).
Thank you again for your help.
It would normally get a pulse length of 1/120 second from half-wave-rectified 60Hz mains. If you don't exceed the voltage-time constant it should be OK.

#### modi72

Joined Apr 23, 2023
9
It would normally get a pulse length of 1/120 second from half-wave-rectified 60Hz mains. If you don't exceed the voltage-time constant it should be OK.
Thank you again. That makes sense.

#### modi72

Joined Apr 23, 2023
9
For anyone interested in this mod. I've made some progress but had some set backs too.

Since the motor is usually fed an AC waveform (positive side of 60Hz AC), to minimize any impact I was hoping to continue feeding it the same 1/120 sec positive sin wave and simply increase the time delay between each wave, like so:

Unfortunately, I couldn't get the circuit to work, so I settled on feeding the motor a DC square wave, as Ian0 suggested, using the following circuit (my apologies about the schematic layout, I'm a noob, so I'm probably breaking some drawing conventions).

Pot P1 and P2 were adjusted so that the motor turned on for 6ms and off for 10ms with P3 at 0 Ohm. This achieved the same coffee flow speed as my unmodified machine. When P3 was turned up to 100K, the 'motor off' period went way up, slowing down the pump, while the 'motor on' period remained 6ms.

I realized that suddenly cutting off the DC supply to the motor would cause back emf, so I placed D6, to help. Or so I thought...

The circuit worked well and I had good flow control down to a very slow flow. Also on full flow, the motor did not overheat .

Unfortunately after a couple of weeks, the machine's mainboard burned out. I'm guessing the back emf did it. I got a new board and reversed all modifications. I feel gun shy about trying anything again

I'm still curious though. So I'm back to my original idea of using a sine wave, increasing the delay between each wave, to slow the pump. I wanted to turn the wave on and off at Zero crossing to eliminate the back emf issue. The circuit I had initially tried was as follows, but it did not work. It used a PANASONIC AQG22105 (SSR RELAY SPST-NO 2A 75-264V) (a zero crossing SSR). I never figured what I did wrong.

If anyone has any input/insight on either circuit please let me know.

#### modi72

Joined Apr 23, 2023
9
For anyone interested in this mod. I've made some progress but had some set backs too.

Since the motor is usually fed an AC waveform (positive side of 60Hz AC), to minimize any impact I was hoping to continue feeding it the same 1/120 sec positive sin wave and simply increase the time delay between each wave, like so:
View attachment 300660

Unfortunately, I couldn't get the circuit to work, so I settled on feeding the motor a DC square wave, as Ian0 suggested, using the following circuit (my apologies about the schematic layout, I'm a noob, so I'm probably breaking some drawing conventions).

View attachment 300659
Pot P1 and P2 were adjusted so that the motor turned on for 6ms and off for 10ms with P3 at 0 Ohm. This achieved the same coffee flow speed as my unmodified machine. When P3 was turned up to 100K, the 'motor off' period went way up, slowing down the pump, while the 'motor on' period remained 6ms.

I realized that suddenly cutting off the DC supply to the motor would cause back emf, so I placed D6, to help. Or so I thought...

The circuit worked well and I had good flow control down to a very slow flow. Also on full flow, the motor did not overheat .

Unfortunately after a couple of weeks, the machine's mainboard burned out. I'm guessing the back emf did it. I got a new board and reversed all modifications. I feel gun shy about trying anything again

I'm still curious though. So I'm back to my original idea of using a sine wave, increasing the delay between each wave, to slow the pump. I wanted to turn the wave on and off at Zero crossing to eliminate the back emf issue. The circuit I had initially tried was as follows, but it did not work. It used a PANASONIC AQG22105 (SSR RELAY SPST-NO 2A 75-264V) (a zero crossing SSR). I never figured what I did wrong.
View attachment 300662

If anyone has any input/insight on either circuit please let me know.
Sorry correction. The SSR was 5V. The circuit was like this

#### Ian0

Joined Aug 7, 2020
8,409
Circuit 1: Unwise choice of value for R6! Otherwise it would have worked.
Replace it by a short.
The diode in series with the motor is superfluous (unless it is built into the motor)

Circuit 2: Too much current in the opto.
Is that a zero-crossing opto-triac? I'm not sure how well it will work as the oscillator phase changes with respect to the mains phase.

#### modi72

Joined Apr 23, 2023
9
Circuit 1: Unwise choice of value for R6! Otherwise it would have worked.
Replace it by a short.
The diode in series with the motor is superfluous (unless it is built into the motor)

Circuit 2: Too much current in the opto.
Is that a zero-crossing opto-triac? I'm not sure how well it will work as the oscillator phase changes with respect to the mains phase.
Thank you Ian0.
Circuit 1. You are correct the resistor in series with the motor is internal to the motor. Regarding R6 (sorry my image might have been too small, I assume you're referring to R4 - 1MOhm which is in series with the flyback diode D6), it was meant to decrease the current caused by the emf. If you have time, can you explain why removing it would be best?

Circuit 2. Yes it is a zero-crossing opto-triac rated at 2A on the load side. The motor draws 0.8A. On the input side, the current should be between 11 to 17mA. As it is drawn, I have 100mA prior to the Zener. The SSR sees 5.1V and it has a 300 Ohms resistance so I should get 16mA through it, and 84mA should be shunted through the Zener.

#### modi72

Joined Apr 23, 2023
9
With respect to AC phase in circuit 2, I turned P1 up to ~15ms, ie just a bit less than 1/60s (so the 'motor on' duration would be 1 single positive wave). As P3 is increased, the duration of the "off time" gradually increases but the duration of time the motor is off does not continually change, it jumps in increments of 1/60s. At least that's what's supposed to happen, if I did my thinking right.

To test my setup I used a 40W incandescent bulb instead of the 40W motor (testing the diming of the bulb here, not for checking EMF obviously). I gradually turned the AC power from 0 to 120V. The bulb would turn on at low voltages, but as soon as the voltage was >50~70VAC, the bulb would turn off ... I couldn't understand why, so I moved on to the DC version.

#### modi72

Joined Apr 23, 2023
9
Thank you Ian0.
Circuit 1. You are correct the resistor in series with the motor is internal to the motor. Regarding R6 (sorry my image might have been too small, I assume you're referring to R4 - 1MOhm which is in series with the flyback diode D6), it was meant to decrease the current caused by the emf. If you have time, can you explain why removing it would be best?

Circuit 2. Yes it is a zero-crossing opto-triac rated at 2A on the load side. The motor draws 0.8A. On the input side, the current should be between 11 to 17mA. As it is drawn, I have 100mA prior to the Zener. The SSR sees 5.1V and it has a 300 Ohms resistance so I should get 16mA through it, and 84mA should be shunted through the Zener.
To answer my own question, the flyback or freewheel diode feeds the back EMF back into the motor (or inductive load) which dissipates the energy, instead of causing a voltage surge. The 1M resistor in series with this diode was preventing the current to flow back through that feedback loop, defeating its function. To quote wikipedia: " A resistor in series with the diode will make the circulating current decay faster at the drawback of an increased reverse voltage. " So with 1MOhm resistor in series, the diode was rendered useless, as Ian0 correctly pointed out !

#### modi72

Joined Apr 23, 2023
9
I made the above SSR circuit work. Because moving the pot makes the period of the on/off cycle vary, there is no way to synchronize this cycling to the 60Hz the motor gets. So, although on average the motor will activate at the desired speed as set by the pot, there's some variability in the frequency (depending on how well the 60HZ AC lines up with the on & off cycle if the 555).

So I went back to the drawing board, and came up with this circuit.

The period of the 555 is fixed ~450ms. The duration of the on cycle varies from 0ms to 450ms (the 10K trims are only there if I find I need a bit of fine tuning). When "on", the zero crossing SSR lets through the normal AC supply, and when off (for the rest of the 450ms cycle) it keeps the AC circuit open.
So when the pot is midway, the motor should pump at its normal rate for 225ms, and rest for 225ms. That way I can gradual adjust the coffee flow from 0 to 100% (depending on the duration of the "on" cycle per 450ms period). Because I'm using a zero crossing SSR, the back emf should be no different than if my circuit were not present.

I tested the circuit with a light bulb instead of the motor. It's working well. The 10K trims were not needed. I just need to workup the nerve to hook it up to the machine...

If anyone wants to replicate this circuit I would suggest choosing a 555 and SSR that both work at 15VDC that way you can remove R1, R5 and the Zener. The 10 K trims also seem unnecessary.
So basically this (which is even more simple):