OK, sorry.

Although virtually all contemporary commercial power supplies will be based on a variation of PWM, by design, they will limit the range of timings, and include filtering, to ensure the output has only a small residue ripple from the switching action, superimposed upon a 'solid' DC level. For the very cheap, switch mode modules available from the usual Internet vendors, that ripple might typically be about 50mV, and for a reputable source like a Siglent, I would expect any ripple to be orders of magnitude lower. Hence, if I had known you were using such a source, I would not have mentioned it.

By contrast, when someone is programming a microcontroller like a Teensy, they may pick virtually any timing values, and often do not include any hardware filtering, so unless I know exactly what the hardware and software is, I put it on the principal suspect's list, until proved innocent.

--------------

Referring back to your previous message, you said "Also the gauge itself uses +15V, -15V and +/- as a reference voltage (+/- referring to the connection between the two power supplies). Could this be a problem related to the amperage issue you referred to above?"

I wasn't sure what you meant by that. I agree that the current demand values on the pair of power supplies looks very suspicious, but I am less clear what is causing it.

Unfortunately, I don't know anything about the internal circuit of the tachometer itself. I am not even clear as to why there are two pointers. The aircraft industry during the period of roughly 1900 to 1980 devised some ingenious electro-mechanical instruments to meet the requirements of the various tasks, which are unique to that industry. But, as I have never used or designed instruments from that period, I am struggling to know if there are any particular points I should be aware of. For now, I am assuming it is essentially a 'disguised' voltmeter for the range -10V to + 10V, with a fairly low input impedance (say 200 Ohm), and obviously scaled in engine RPM, rather than volts - but this may be far too simplistic. Any advice or pointers to relevant information could be helpful.

Best wishes, Dave

Thanks for taking the time to respond. Yes, the end result will use the Teensy to power the signal source side of the circuit. But for testing, I am using a bench DC power supply which, I’m fairly confident, does not rely on PWM generated voltage. I will double check that (it’s a Siglent 3300). 

In the meantime, I tried it again today and oddly it is significantly smoother. Not 100% perfect, but about 95% there. Hooked it up to the Teensy and it works similarly.

Jon

   I assume you are using a PWM output from Teensy as the source for the variable voltage input to your circuit, as in your opening message you said " The controller has the ability to supply PWM 3.3v signals at 10mA max.", and continued to describe the unit you needed to drive, etc..

PWM stands for Pulse Width Modulation. This is a technique for producing 'variable' voltages from a digital circuit.

Simplistically. for a processor like the ESP32, a PWM output, (provided it is not connected to a low impedance load,) will be at (nearly) 3.3V or (nearly) 0V, at any given time. To produce "1.65V", it will repeatedly output 3.3V for a short time, and then 0V for the same short time. Hence, the "average" voltage will ((0 + 3.3)/2) = 1.65V.  Thus, it is Pulse Width Modulated, because the wider the "On" time, compared to the "Off" time, then the nearer the average voltage will be to 3.3V.

If the device being driven by this waveform has sufficient 'inertia', such then it will only show the 'average' result, then the result will be as required, but if the device can 'follow' the pulses, then it will display the pulsed nature. The 'inertia' can be 'real', if the device is a mechanical device such as a solenoid, or 'virtual', such as an LED.

For example, PWM is often used as a dimmer control on LEDs, where the power is switched at (say) 10kHz. To a human eye, the 'vision persistence' means we can only see individual flashes up to a rate of (say) 20 times per second, so it appears the LED is constantly lit, but a high speed video camera can capture it switching on and off.

I am  sure the Internet will have many articles describing PWM in more detail ... I noticed this one gave a simple introduction:

https://www.pjrc.com/teensy/td_pulse.html

In addition Bill (@dronebot-workshop) has discussed PWM on a number of occasions, including this article on driving motors:

https://dronebotworkshop.com/dc-gearmotors-pwm/

Note, it should now become obvious that if you have access to an oscilloscope, then examining the Teensy output signal could be very useful to help you decide how to proceed!

----------------------

As I do not know the parameters (time/frequency) of the PWM from your Teensy board, which will be largely defined in software, I do not know whether your 'output device' is trying to follow some of the pulses. Hence, my first suggestion is to test your circuit that is 'definitely' a variable DC voltage, not a PWM waveform.

I do not know whether your 'output device' will continue to twitch and jerk with the DC source, (because this is not the problem), or will become steady. But as with all the best detective stories, it is a case of examining each suspect in turn, and eliminating or incriminating them on the basis of evidence. This is my first suspect ... it may not be my last ... but I'll leave the others for another installment, if required!

I hope this makes a little sense.  I am sorry you are meeting a lot of new concepts, and this can be hard to begin with, but they will all become familiar in time.

Best wishes, Dave

Thanks for the response. I am using a separate power source to provide the signal (the white power supply in the video), so it is not the Teensy. What would be the next step? I am going to build the schematic above (the one with the transistors and the three op amps) and will see how that compares.

As always, MANY thanks!

Jon

Good to see you are making progress. 

I am not clear of the source of the jitter, so I recommend a systematic approach to discovering the cause.

I think my first attempt to discovering the cause would be to provide a temporary variable DC source of 0 to 3.3V to feed to Vin, instead of the Teensy, to see if it still jittered. My reasoning being the Teensy output will be a PWM signal, not DC, and MIGHT be the origin.

Remember, the op-amp circuit above will have an input impedance of 10k ... that is, if you arrange a variable voltage using (say) resistors and a potentiometer, then when the circuit is operating, then the output of the potentiometer will apparently be connecting via a 10k resistor, to ground, which will result in a lower voltage output from the pot, compared to when the pot output is unconnected.

If there is no jitter with the DC source, then an RC filter between the Teensy output and Vin (or possibly between U1B and U1c) might help. To assist with designing this, it would be helpful to know the waveform timing from the Teensy, indicating the likely range of waveforms from minimum to maximum, whilst avoiding the theoretical extremes of always on(high) and always off(low), that are obviously DC.

However, if it jitters with a DC source, then other problems need to be considered.

Best wishes, Dave

First, allow me to say how thankful I am for your willingness to help me out and your generosity of knowledge. For someone new to this as I am, it mean more than you know!

While waiting for the correct transistors to arrive (Dave - I got the MJ2955 coming from Mouser tomorrow), I constructed the circuit that @bbutcher85 proposed on 8/1 (scroll back) which I will continue to refine using both of your suggestions (esp with the push/pull transistors). I was indeed able to get a 0-3.3vDC signal to modulate a -10 to +10vDC signal.

I hooked it up into the gauge that I am trying to power (N1 tachometer which is the primary indication of how fast the big fan of a jet engine is turning) and got it to work but with some serious jitteriness. Here is a video I took to give you some idea of what's going on..

https://youtu.be/9PNtLochfrs

I installed 0.1uF capacitors between the V+ and V- (rails) and ground (+/-) without any luck. Also the gauge itself uses +15V, -15V and +/- as a reference voltage (+/- referring to the connection between the two power supplies). Could this be a problem related to the amperage issue you referred to above?

THANK YOU!

Jon

  re: I was able to locate the 2N3055 bipolar transistor, but was unable to find the 2N2955 transistor...am I missing something? Also, what would you recommend for R8?

As the circuit you were referring to was designed by @bbutcher85, I was respectfully waiting for them to reply with definitive answers, which will probably happen soon.

In the meantime, I'll venture my suggestions. Please note, I have not tried or analysed the circuit suggested, so these answers are just based on my first thoughts, based on the devices in general usage.

--------------

The 2N2035 and 2N2955 transistors are NPN and PNP complementary 'equivalents' of each other that originated over 50 years ago. The 2N2035 (NPN) appeared in many more circuits than the 2N2955 (PNP), and hence is probably easier to find. The 2N2035 in particular, became 'famous', as the 'go-to' device for fairly high power circuits, but was originally linked to the relatively inconvenient metal package.

Hence, some semiconductor manufacturers produced "clones" of a similar device in a plastic package, with '***2035' style type numbers, where '***' was replaced by two or three letters, often loosely based on the manufacturer's name.

Of course, these "clones" have an entirely different physical appearance, and may have had different electrical characteristics, but in most circuits, would probably perform in a similar manner to the original 2N3055. (It would be the responsibility of the circuit designer to ensure equivalence in a production scenario, by carefully checking the data sheets, and then testing pre-production samples, as there could be 'unexpected' issues.)

For example, the metal can device was credited with 115W power dissipation capability, whilst I have seen 90W quoted for one of the plastic devices. In practice, it needs a very substantial heatsink (at least without fans or other add-ons) to dissipate 90W, and most circuits, including yours, will operate well below this level.

I haven't done a proper search or comparison of devices, but TIP (TIP2035) and MJ (MJ2035) seem popular choices.

And you will not be surprised to find, TIP2955 and MJ2955, the PNP version, are also listed, which may be easier for you to source.

However, confusingly, some of the TIP and MJ devices are also listed in the metal can packaging, so check before you buy!

------------------

With op-amps in the inverting gain configuration of U1B, it is common to choose the two input resistors to be the same value. Hence, my first guess is R8 would be the same value as R6. R6 is 10kOhm.

-----------------

As I started, @bbutcher85 may respond with more authoritative answers. I am only trying to fill a gap, and may have missed something.

Hope this helps. Best wishes, Dave

THANK YOU so very much for your input on this. I have learned a tremendous amount just from your discussion about this. I'm sorry I cant provide more technical information about the amperage requirements of or the impedance of the load that this is going to. Once I get this up and running I'm sure that I will have that information more readily. I have ordered all fo the components from Mouser outlined in the first schematic and am going to order the components in the second schematic. I was able to locate the 2N3055 bipolar transistor, but was unable to find the 2N2955 transistor...am I missing something? Also, what would you recommend for R8?

Jon

The attached circuit has the same features as the original, but adds a class B push-pull output capable of delivering high currents in both polarities. Class B push-pull amplifiers do have a problem with distortion of the signal since near zero output neither the NPN or the PNP transistor are biased on over about a +/- 0.7 volt range. The third op amp stage provides feedback to minimize the distortion of the signal. 

Class A push-pull amplifiers are another option, but they tend to consume a lot of power and create a lot of heat.

There is a lot of information available on the internet regarding op amps and push-pull amplifiers for anyone interested in the subject.