Do I need some sort of buffer?

Question

I'm using a CA3080 op amp in a transimpedance configuration to convert the current flowing through a phototube into a voltage that I can display on an oscilloscope.

enter image description here

With the output fed straight into an oscilloscope, there's no problem; it works great showing positive DC values with a constant light source or signals like the IR pulses coming from a DVR remote control. I have also tried connecting the transimpedance amplifier to an audio amplifier to make the signals audible and that works. What I would like to do, though, is have both an oscilloscope and an audio amplifier connected at the same time; when I do this, the signal shown by the oscilloscope collapses; I can turn the vertical gain up on the scope but the signal is very noisy. Do I need to add some sort of buffer amplifier that I can put between the transimpedance amplifier output and both the scope and audio amp inputs so that the scope input isn't loaded down by the audio amplifier? I have 1458, TL072, and CA3080 op amps on hand.

Also: The scope inputs are 1M/13pF and it's set for DC coupling because I want the scope to show +DC levels. The audio amp in question I'm not so sure about. It's got a 47k in series with a 0.01uF to ground and a 0.001uF straight to ground. From there it goes to one leg of a 500k input level pot with the other leg to ground; the wiper goes right into the grid of a 12AU7A triode tube. The position of the input pot doesn't make a difference.

• Drew: I'd add a second buffer between the trans-impedance amp and the audio amp. The fact that is loading down the output so much seems like a bad sign.
• Whit3rd: CA3080 is not an op amp (doesn't have low-impedance output) but an operational transconductance amplifier. It has output current source, the load on that output sets the voltage gain. It cannot be left without an output load, so ought not to be 'buffered' by the usual op amp circuit.
• WatcherOfAll: @Whit3rd How would it be to throw a resistor between the CA3080 and ground to give it a consistent load? BTW right now I've got another CA3080 set up as a buffer (output connected to inv input) and a capacitor between its output and the audio amp (the buffer output is always positive). The scope is connected to the t-i amp output and buffer input and it looks fine. I have tried a 1458 and a TL072 as a buffer and they just don't work at all. I can run like I am but the audio amp has to be cranked to get decent volume from the speaker.
• Whit3rd: An op amp with resistor from output to (-) and grounded (+) is a good load (inverting buffer) for a CA3080. With program current 1 mA, a 1k feedback resistor gets you 1V scale, and the (-) input of the op amp stays pseudo-grounded. The (later design) LM13700 has two 3080-like amplifiers and darlington transistors for buffers, might be easier to work with. Pseudoground is the ideal 'load' for a current source.
• Lundin: Note that these ancient parts from the previous millennium will have non-existent ESD protection. So always consider the situation where the part is just dead because of that.

Answer 1

Cool, the CA3080 is from 1969! That's a historically important IC, as it was the first operational transconductance amplifier (OTA) IC. Probably not the easiest OTA (as said, not a simple opamp) to work with, comparatively.

As noted by Whit3rd, an early but useful OTA (if a bit noisy, maybe, compared to modern alternatives), but unlike an opamp, its output is a well-defined current, not voltage. So, you cannot just load the output with multiple different loads and look at voltages across the loads.

Attaching you attaching your 1 MΩ scope lead to a conversion of 1 μA of output current into a voltage of 1 V. Clearly, attaching a much-less-impedance, and more importantly, frequency-selective second sink for current will disrupt that!

So, bit of a crossroads: You can build transimpedance amplifiers out of "regular" opamps, and by choosing a modern opamp with sufficient drive strength, you might avoid the immediate visible problems (part number is just an example; main point here is to use a somewhat modern opamp with low bias current).

TIA circuit
^{Transimpedance amplifier for a photodetector. From: Linear Technology, Transimpedance Amplifiers for Wide Range Photodiodes Have Challenging Requirements by Brian Black and Glen Brisebois. Available online}

However, by directly loading the output of this circuit, you'd be prone to affecting the low-pass behaviour in hard-to-predict ways, so I'd always add a simple buffer opamp - not a problem, because many (most?) opamps in packages you want to deal with come in multiple-opamps-per-package, so you'd always have at least two at hand. Also allows you to invert the voltage if you want! Honestly, I'd go as far as saying that you should have two output buffers, one for each your oscilloscope and the lower-impedance amplifier, to definitely decouple them.

You'd deliberately choose a C_F that's larger than the parasitic capacitance, to limit the bandwidth, hence increasing stability and decreasing noise.

I have 1458, TL072, and CA3080 op amps on hand.

Uh, you seem to live in a time capsule full of ICs that are without exception older than 50 years!

• The {MC,LM,...}1458 is just two μA741 in a package – the pretty much worst opamp you'll find. There's a whole lot of explanations why you wouldn't want to use that.

  So, wouldn't use that. Don't think that thing would have a sensible use case.

• The CA3080 from 1969, though generally useful, is really the first generation of OTAs. Noisy, and lacking the bias current input which allows for compensation. In use cases where you really need an OTA, you'd want an OTA that has such (and probably lower noise).

• The TL072 is from the mid-1970s (so, didn't do the "jay we now know how to do integrated circuits" first-hour mistakes that the other two did). It's a PFET input opamp – so, it does have low bias current. Problem with the TL072 is the rather large supply voltage you need and the limited DC accuracy.

  Of these three, it's your best choice, but, hm, a good choice would be a low-bias-, low-offset-current input, opamp with sensible band limitation (you seem to be interested in audio frequency range, so why allow noise from MHz of bandwidth to seep into your system), less necessary voltage supply headroom and good temperature and power supply noise behaviour; say, for example, the OPA396.

  But, overall, the TL072 would be a good choice, if you can make sure things don't get too hot (the offset current can get large on the TL072, which becomes a problem in transimpedance amplfiers; the OPA396 would be about a 1000 times better in that respect).

Note that the physical design of the feedback loop in any current amplifier is important if the input currents are so small – anything you pick up inductively (say, because your resistor gets connected with wires that effectively form a loop that forms the secondary side of a transformer with your surrounding mains installation wiring) is going to be visible in the output. If you care about accuracy, make sure you buy resistors that have good accuracy, too – state of the art low-cost would be 0.1% tolerance.

• periblepsis: Nice!! Some of my thoughts are captured here. But much more, as well. My focus is always on the physics, so my mind went directly to the photodiode itself and then straight to Hamamatsu where Figure 2 expresses my first thought of a single voltage supply circuit. My next thought when straight to your own thoughts of "attaching you attaching your 1 MΩ scope lead..." and "attaching a much-less-frequency-selective impedance". Nail on!
• Marcus Müller: @periblepsis nice catalogue, thanks!