2
\$\begingroup\$

Background

A while ago I got my dad's old Musical Fidelity A1 MKI integrated amplifier from the early 80s (I believe). I found a blog that featured a lot of interesting information about this amplifier (Musical Fidelity A1, Mark Hennessy), and in this blog, the author suggest building a new preamp if you're having problems with the volume pot and input selector switch instead of just replacing them and facing the same problems a couple of months later. I knew from replacing the electrolytic capacitors that there was some oxidation in the amplifier, and the input selector switch sometimes only lets through one channel. Therefore, I figured I could try building this preamp since there seems to be improvements to the quality of sound and tracking of the channels.

Circuit

On the following page (Modifications), all the information about the preamp circuit can be found, but you can also see the schematic here (all props to Mark Hennessy): Preamp circuit

I struggled to find the NE5532 and OPA2134 in a DIP package under 8$ (!!!) here in Sweden, and since the author says that you have a lot of freedom in choosing and opamp (with some modifications to the circuit, I'll get to that later) I thought that I could find another opamp to use in this circuit. When using another opamp, the author suggest changing the values of R3 and R4 to 3k and 1k ohms respectively, which I have done. The values of C3 and C4 are also changed to 100nF. After some research online, I saw several people mentioning the LM4562 opamp, and when I compared the datasheets of the OPA2134 and LM4562, they were quite similar except a few parameters, so I bought that one and used it in the circuit. Worth mentioning that I built the circuit on stripboard in the exact same way Mark did (e.g. same layout, making the strips as short as possible, etc) (2). Mark also warned about high-speed opamps on the modification page, but I'll get to that later as well.

Problem

Now to the problem I am facing. When testing the circuit with a power supply, function generator and an oscilloscope, I get an excessive, fluctuating DC offset on the outputs of the opamps. The circuit is still technically "working", and by that I mean that if you have a sinusoidal input, the output is also a nice looking sinusoidal curve but amplified (see picture below)(Unfortunately, I didn't have enough probes to show the input, but you'll just have to trust me that this is the amplified signal ;)): enter image description here

Because of this, I think we can rule out any obvious mistakes when making the circuit on the stripboard. The DC offset on one of the channels is not that bad (max -0.5V, typ 0.05V), but still "too large" if you look at the datasheet of the LM4562 (LM4562). The other channel on other hand had HUGE DC offsets. When powering the circuit for the first time yesterday I had a whopping -11V on the output!! This later got smaller, slowly becoming less and less (-9V, -6V then stabilizing at around -2V). When testing the preamp today, it started at -6V but then got smaller and again stabilizing at -2V. However, sometimes the offset jumps down to practically nothing, and sometimes it likes to stay at -0.8V, as in the picture above (10x probe). It could be that these offsets are no problem since the signal has to pass through 1uF polyester capacitors before entering the power amp, but I don't like using a circuit that doesn't behave as expected even though it might work in practice. I'm taking no chances here.

As I quickly mentioned above, the author of the blog warns us about using a high-speed opamp in this circuit in the modification page, but only having learnt ideal opamps so far in school, and since I thought the parameters of the OPA2134 and LM4562 were similar enough, and since I thought there was "enough" filtering capacitors in this circuit, I still went with that opamp. I also tried the same amplifier circuit on one of the channels on the TL084, and it worked just like expected. I don't know whether or not it is because of the higher slew-rate or GBWP of the LM4562, but I think it's one of these two that might cause the problems.

I found a discussion on another forum where there seems to be a person having the same problems as me. I read it, but I don't think I understood everything that was said. However I tried following some of the suggestions found there and here is what happend. Firstly I tried rotating the pot, and just like him nothing happened to the DC offset when doing so. I also tried putting a 1k resistor in parallell to R2, and just like him the DC offset got smaller, but the noise on the output got really bad.

My Question In the configuration described above, do you see any improvements/changes/additions that could be made to make the DC offset go back to normal? I could just buy another opamp since the circuit worked well with the TL084, but people really seems to like the sound of the LM4562, so it would be fun getting this circuit to behave as expected. It's also always fun to learn something new. Thanks in advance for any answers!

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Check R2 connections. Most likely an intermittent there. \$\endgroup\$ Commented Mar 28 at 17:27
  • \$\begingroup\$ Great question... \$\endgroup\$
    – Voltage Spike
    Commented Mar 28 at 17:54
  • \$\begingroup\$ You might try increasing the values of R3 and R4 to make their parallel combination the same as R2 (68K). The device has a differential input impedance of only 30K. \$\endgroup\$ Commented Mar 28 at 19:23
  • \$\begingroup\$ @JohnBirckhead I've checked them and they seem fine, but I can try resolder them as well. However, R2 on the side that has the highest DC offset spans one columns further than on the other side, could that cause any trouble do you think? What does the differential input impedance tell me when designing a circuit? If the difference in impedance is too large, is that always going to cause trouble? I also had to mount R4 on the side with biggest DC offset in a vertical position since it didn't fit otherwise, and therefore that path is longer than on the other side, could that cause any problems? \$\endgroup\$ Commented Mar 29 at 12:38

2 Answers 2

1
\$\begingroup\$

Maybe you have very high frequency oscillations at the output that you aren't seeing on the oscilloscope due to aliasing, bandwidth limit, or low probe bandwidth.

That could cause signal asymmetry and clipping that might show up on the 'scope as a DC offset, or other low frequency component in the audio band.

You could try deliberately lowering the bandwidth of the amplifier, by placing a 1nF or so capacitor in parallel with R3.

Also, you don't say what you did with the other op-amp in the package. Don't leave its inputs unconnected, because that's a sure-fire way to make oscillations and pops. Since you have dual supplies, here's an easy way to keep the other op-amp quiet:

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
5
  • \$\begingroup\$ Thank you for your answer! Unfortunately I won't have access to the testing equipment for about a week, but I'll try putting a capacitor in parallel with R3 as soon as I can. The other channel in the opamp is connected to the other side of the circuit since it's soldered to the stripboard, so it's not unconnected. I can try sending the same input through both sides of the circuit and see if that changes anything. \$\endgroup\$ Commented Mar 29 at 12:05
  • \$\begingroup\$ @jonatan.wilund It's not sufficient to solder a pin to a strip, to call it "connected". The pins must be explicitly held at some non-floating potential. I've added a suggestion for this. \$\endgroup\$ Commented Mar 29 at 12:40
  • \$\begingroup\$ Alright! Since I tested the same circuit on one of the TL084 channels and didn't hold the rest of the channels inputs at a specific voltage (they were not connected to anything) and that worked fine, would you say it is more important to do so when using a high speed opamp? \$\endgroup\$ Commented Mar 29 at 12:49
  • \$\begingroup\$ @jonatan.wilund I would say it's \$\frac{55MHz}{3MHz}\$ times more important than the already very important! I would also say you got lucky with the TL084, if you left 3 op-amps with inputs floating \$\endgroup\$ Commented Mar 29 at 13:01
  • \$\begingroup\$ Okay, I'll definitely try testing this circuit again while holding the other opamp's non-inverting input at 0V. Thank you again for your answers Simon! \$\endgroup\$ Commented Mar 29 at 13:35
1
\$\begingroup\$

I cannot say why so high offset, and don´t understand why connecting 1k resistor in parallel with R2 has influence in offset and noise. Anyhow, you can reduce DC gain in this non inverting amplifier blocking DC component with a capacitor, in the path of R4 to ground. DC gain will be reduce from current 4.09 to 1.

R3 and R4 should be increase for better balance and also helps to reduce capacitor value. The impedance of a 10 uF capacitor at 20 Hz is about 800 Ohm, so gain at that frecuency will be 3.98, about 3 % lower. You can increase the capcitor value if better bandwith is desired at low frecuency; for higher frecuency, the gain will remain as current setup.

enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ Thanks, I like the idea of increasing the R3 and R4. My guess is that when connecting a 1k resistor in parallel to R2, the resulting resistance is approx 1k ohm which is much closer to 4k which I have on the output. Since the resistance is relatively low, this might be the reason for the increased noise, but I'm just guessing here. I'll try increasing R3 and R4 when I get home again, and if that doesn't work I'll definitely try the capacitor trick, very clever! \$\endgroup\$ Commented Mar 30 at 9:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.