Building an OPA541-based power amplifier for function generators

แชร์
ฝัง
  • เผยแพร่เมื่อ 1 ธ.ค. 2024

ความคิดเห็น • 41

  • @antilogism
    @antilogism 2 ปีที่แล้ว +1

    I just "finished" a similar multipurpose 541-based bench buffer yesterday---and then I see this video...Dash it all! Like yours, my board already had the DC coupling as well. The 541 seems to be an old-stock Burr-Brown part.
    My goal was just +/- 10 V (20 V p-p) at only 4 Amperes so I used a couple 15 V switchers. My hacky cooling used a slow, quiet, 24 V, 60 mm fan with a 220 Ω series resistor across the 30 Volts. I beat the tar out of it yesterday, running transformers, motors, near-shorts and other evil loads with a function generator and bench supply and the cooling held. The input offset was way lower than expected at much less than 1 mV. The rms noise, taken with a wideband thermal rms meter, was less than 2 mV (unloaded @ 10Vdc) and some of that could be from the linear bench supply used as a reference source. Next is to add monitor BNCs for current and voltage to make it easy to connect a scope for power measurements with reactive loads.
    Edit:
    On the glitch at 23:45, I see that too and not just at higher frequencies. I've been trying to understand it. It only happens when the signal crosses zero. A +2.5/-2.5 square-wave shows it clearly but a 0 to +5 V or 0 to -5 V square is fine. It looks like some form of crossover distortion but only when loaded heavily. Counterfeit chip?
    Yet another edit:
    Confirmed: Crossover distortion. One of those cases where textbook art and a real scope trace diverge. Sadly, seems normal for this chip at high currents. See TI thread "opa541-distortion-at-the-bottom-of-output-waveform".

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว

      Thanks for your comments and experiences with your build. Very useful stuff. BTW, I would not recommend using panel meters like I did. They are too slow. A better way may be an (averaging) watt meter in the output but that would probably be be a self-build. I do like the capability of adjusting the operating voltage but at the moment this depends on having a VARIAC at hand. Its fine for me but not a general solution.

    • @antilogism
      @antilogism 2 ปีที่แล้ว +2

      ​@@TheHWcave Mine is maybe a bit more spartan and not nearly as polished as yours. My labels are by Sharpie pen! My box is pretty small too so not much room left.
      Today I squeezed in a current circuit. A salvaged Tamura S22P current sensor, and a 7805 for it's power, amplified by an LF353 dual op-amp into the box on a little perf-board. That assembly is biased to get 1 V/A. That Tamura supposedly works to 100 kHz, which was the design goal for my output. That circuit leads to a BNC on the front panel that I patch into the scope set to 1 A to 1 V. With volts on the other channel I get both components, in time, for watts. Using the product of the two channels in the scope's Math, I get power at any point. Averaging the product gives Watts over the sweep period. Surprisingly, the scope is smart enough to label the math trace "Watts"! I calibrated at DC and it measures well, from what I can tell.

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว +1

      @@antilogism Thanks for that tip on the Tamura sensor. That may come useful once I try to build my own power meter.

    • @antilogism
      @antilogism 2 ปีที่แล้ว

      LEM has another similar line. Also Allegro has some nice chips and several evaluation boards. They have at least one hybrid chip with voltage and current, including an integrated ADC, that crunches watts and uses I2C.

  • @plainedgedsaw1694
    @plainedgedsaw1694 3 ปีที่แล้ว +1

    Nice build.

  • @BjornV78
    @BjornV78 3 ปีที่แล้ว +1

    Nice and clean build.
    The white wireconnector blocks next to the rotary switch, i presume that those are glued to the frontpanel ?

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว +1

      Yes, glue was the easiest solution. If you take care in cleaning the surfaces first and flattening the terminal-block underside it works quite well.

    • @BjornV78
      @BjornV78 3 ปีที่แล้ว +1

      @@TheHWcave another solution (but not so quick) is to drill 1 or 2 small holes through the frontplate and small bolts and countersink the heads in the frontplate and use nuts to lock the connectorblocks in place. Then finally hide the bolt heads behind the coverplate foil. But glueing is definitely way easier.

  • @meowme7644
    @meowme7644 2 ปีที่แล้ว +1

    👍👍👍 thx

  • @hullinstruments
    @hullinstruments ปีที่แล้ว +1

    Have you seen the amps sold by feeltech/feelelect. Made specifically for their function generators. They look pretty interesting. And offer different versions capable of a wide range of power output.

    • @TheHWcave
      @TheHWcave  ปีที่แล้ว

      i had not looked for a while. Interesting, Thanks for pointing them out.

    • @hullinstruments
      @hullinstruments ปีที่แล้ว

      @@TheHWcave let k ow what you think
      I'll be interested in hearing your opinion

    • @TheHWcave
      @TheHWcave  ปีที่แล้ว

      ​@@hullinstruments I have to disappoint you because I have no time for an in-depth tour of available amps at the moment. Sorry.

  • @nityanandadas5575
    @nityanandadas5575 ปีที่แล้ว

    Excellent.
    Can this amplifier drive resistive load of 1 Ω ? Current required is 100 mA .
    Actually I need the output impedance of the amplifier to be less than 1Ω .
    Please.

  • @lukasz_kostka
    @lukasz_kostka ปีที่แล้ว

    Where did you connect gain control knob to amp ?

    • @TheHWcave
      @TheHWcave  ปีที่แล้ว

      My gain control sits in between the front panel input connector and the OPA541 module input connector (screw terminal). The OPA541 is running at its default x33 setting.

  • @antilogism
    @antilogism 2 ปีที่แล้ว +1

    The OPA541 has very heavy compensation (read that a crazy slow slew rate) but running it at unity gain helps with that.

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว +1

      True, but I was to lazy to change the gain on the board... I am using it mostly for DC to audio frequencies and predominantly sine curves and its doing just fine.

  • @sharanyas-ve2fy
    @sharanyas-ve2fy ปีที่แล้ว

    Its an interesting work.I would like to know whether i can use this amplifier to power my Helmholtz coil(4-5 ohm resistance)which needs an AC magnetic field of 40 Gauss for that requires a current of 4 A. will the function generator connected to this circuit able to meet my needs?

    • @TheHWcave
      @TheHWcave  ปีที่แล้ว

      It might, but its really a tricky case maxing the chip and the cooling system pretty much out. You also need to carefully design the supply voltage to something like 22V max (4A*4Ohm = 16V + 5.5V voltage swing). But wouldn't it be easier to feed your coil through a transformer from mains?

    • @sharanyas-ve2fy
      @sharanyas-ve2fy ปีที่แล้ว +1

      @@TheHWcave Thank you for your reply. I want to conduct a study by varying the frequency, i suppose it will be easier to feed the signal through a function generator for that. I tried with an audio amplifier but i dnt get the current required. Your suggestions will be a great help to me

  • @sdriza
    @sdriza 3 ปีที่แล้ว

    Perfect timing...

  • @Steven_Bennett_YT
    @Steven_Bennett_YT 3 ปีที่แล้ว

    Another interesting and useful video - thanks. Do you have a link to the case purchase site please?

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว

      The eBay seller was ukstore829_3
      www.ebay.co.uk/itm/114023382976?hash=item1a8c52bfc0:g:9HkAAOSwjYld7rtE

  • @t1d100
    @t1d100 3 ปีที่แล้ว +2

    Another excellent video! Another excellent build! Kudos! How convinced are you that the OPA-541 module is using genuine OEM components? I have only recently become aware of the 78C12 Switch Mode V Regs. IIRC, when I looked them up on Mouser, they were rather expensive and only came in limited voltage values. I imagine that they will become cheaper and be offered in more voltages, as they become more popular, which they certainly will. However, they might be just the thing, as a solution to some particular design issue. Please, please continue with your developments and corrections of your design and keep us informed. In the meantime, it would be great if you would post, here, a list of its issues and your thoughts about how you might address them. I rather think that this is a project that I need to build. I ran into the same lack of power, recently, testing a circuit.

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว +1

      Thanks for the comments. Are the chips genuine? Hard to say. This was one of the reasons I did the power bandwidth test from the data sheet. As far as I can tell, it behaves as expected and the performance is good enough for me (considering cost). The main flaw with this module is the totally inadequate heat sink and to make matters worse, it appears to be put on with thermal adhesive/epoxy. I spent half a day just trying to remove it...

  • @steve6375
    @steve6375 2 ปีที่แล้ว

    Those meters measure common current in 0V rail (low ohm shunt inside meter on 0V rail). As an experiment, maybe try powering them independently from two 9V batteries (and no fan cct) and see if that current reading goes to zero instead of 40mA with no load?

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว

      An interesting suggestion. I may try that but it will be a while because I am too busy and also I don't want to mess with the amplifier at the moment and take the meters out for experiments.

    • @steve6375
      @steve6375 2 ปีที่แล้ว

      @@TheHWcave ok, but you don't need to remove meters, just detach power lead and connect to 9v battery +ve with negative of battery connected to 0v (reverse connections for -ve meter) .

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว

      @@steve6375 Trust me, I know how packed the wiring is on the front plate where the meters are.

  • @davidkclayton
    @davidkclayton 3 ปีที่แล้ว +1

    Analog current meters would respond faster to changes. Another thought would be to use a switch mode power supplies with a feedback loop to control their output it's not that hard to do

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว +1

      I considered analog meters and even bought two but these are even worse in responding. As it turns out currents are pretty much a mirror image of the output split into + and - components. This means you need true RMS and good frequency bandwidth. In the end I think it is not worth having any current meters in the rails. Maybe a peak current LED could be useful.
      I had not thought about rolling my own SMPS to provide adjustable voltages. That may be something to investigate but for me my VARIAC is way more simpler in adjusting the voltage.

    • @davidkclayton
      @davidkclayton 3 ปีที่แล้ว +1

      @@TheHWcave of course if you like LEDs then check out the LM3915

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว

      @@davidkclayton Maybe something like that. Possibly just one row of LEDs connected to indicate the (rectified) peak output.current

  • @pro.empire
    @pro.empire 2 ปีที่แล้ว

    do you sell this? and you have this at the 40 and up to 45khz?

    • @TheHWcave
      @TheHWcave  2 ปีที่แล้ว

      Sorry, no I am not selling. I built it because I sometimes need the extra voltage or current the amp provides and I have not found anything (affordable) that could replace it.

  • @isidoromaich7226
    @isidoromaich7226 3 ปีที่แล้ว

    Have you seen this? ->th-cam.com/video/4Fb7Eqe0PuE/w-d-xo.html
    It seems that you need to address some issues at the output of this kind of amps

    • @TheHWcave
      @TheHWcave  3 ปีที่แล้ว

      Sounds interesting, I may do some tests on that