How do you calculate/measure the output impedence of a circuit?

[Otbaudio]

Trying to work out the output impedence of a circuit I've put together, but of the methods I've found so far they require an oscilloscope and a signal generator to work it out. All I have is a multimeter haha. Is it possible to work out the output impedence of the circuit with just a multimeter, and if so, how?

Thanks in advance

Oli

[EBK]

Do you have a schematic?

[R.G.]

Yes, it is possible, but not easy to get accurate results.

The process is to run a signal generator into the circuit (oops... need a signal generator of some kind, too!) set the output signal from the device to some convenient level, as measured by your multimeter on its "AC Volts" setting, with no load. You might set it so the meter reads 1.00V AC. Note that your meter does not in general read the actual, correct value on AC volts, just an estimate. But this doesn't matter for your use in this case. With the set level of the output voltage steady on the circuit output, connect a load resistor to the circuit output. The output impedance of the circuit and the added load resistor make a voltage divider, and the voltage you read on the output across the added load resistor will drop some. You can then work out the circuit output impedance by how much the voltage drops.

There are problems with this, of course. You have to have a signal generator. The circuit must block any DC from its output so it doesn't fool your meter. The output loading may have to be severe, and possibly damaging to the circuit if the circuit doesn't have some kind of internal immunity to too-low loads.

In my experience, I haven't usually needed to know exactly what the output impedance is, only a reliable order of magnitude estimate. That is, I need to know if the output impedance is 1M, 100K, 10K, 1K, 100 ohms, or less than 1 ohm. For discrete circuits that do not use feedback, this is easy. The amplifier forms with a load resistor from the collector or drain to the power supply have an output impedance is fairly accurately estimated as the collector/drain resistance. The follower forms, emitter or source follower, are tougher to estimate, as they're effectively the emitter/source resistor divided by the internal gain/workings of the active device. These are in general "low" - in the sub 1K, and sometimes sub-100 ohm range, even for FETs.

Circuits with overall feedback, like opamps, are much harder to get accurate results, but easier to estimate. The feedback tries to force the output to a constant level as long as the internal devices can possibly do it. A normal opamp will have an effective output impedance of well under 1 ohm for loads that don't require too much current. As the delivered output current goes up, the feedback can't cover up the loading as well, and the output will start acting like it has a higher output impedance. Then when the internal current limiting protection is activated, it just clips, changing the output waveform and effectively fooling a multimeter-only setup. So for opamps, just assume the output impedance is "very low", under 100 ohms or so, plus any series resistance you stick in series with the output.

If you need accurate, real results, it's going to cost you a trip to the oscilloscope and signal generator store, plus some more detailed study on what and how output impedance is.

A lot depends on what you need the number for. If it's for your own knowledge, order-of magnitude is usually good enough.

[Otbaudio]

That's really helpful thank you!

Well I'm hoping to sell these on my Reverb store, and just trying to get the specs sorted for customers. Do you think it matters that much? I know all the pedals I've bought before have had an output impedence on the specs but idk if that's just the done thing or if it's a requirement.

Thoughts?

Unfortunately I don't have schematic to hand but after the output capacitor of my circuit I have a current limiting resistor of 220k which sounds lush on its own, but I'm thinking that's probably a bit high for running into another pedal. I just know any lower than 150k gets unbearably harsh on the ears. But tbh i don't really understand how it works, output impedence is one of the things I've just not managed to get my head around!

Oli

[antonis]

<But tbh i don't really understand how it works, output impedence is one of the things I've just not managed to get my head around!>
Just imagine output impedance of an arbitrary circuit as source internal resistance to whatever this very circuit drives...

[Otbaudio]

Okay, how would one measure what that is haha? Bear with me I'm still pretty new at this and not particularly fluent in multimeter use yet.

[R.G.]

To a good first approximation, the output impedance of your circuit is that 220K series resistor. Most solid state circuits have quite a bit lower impedance than that.

You may need to put in some time figuring out why your circuit needs such a big output resistor. The problem is this: that large output resistor means that whatever is connected to the output of the circuit has a big effect on how big the signal that gets into the unknown following device, and probably on its frequency characteristics. Of course, the actual device it's connected to is unknown, so a good and workmanlike approach would be to make your circuit more tolerant of whatever is connected after it. Otherwise, your customers will get a variable result depending on what and how they connect it up.

Many guitar amps have a 1M nominal input resistance. So if you're connected to just such an amp, your circuit will lose 220K/1000K of its output, about 1/5. But most of your customers will connect it to a random assortment of other pedals, some of which will have been made by beginning pedal makers, and have a wildly different input impedance, and so will have wildly different results.

I suggest that you add some loading resistor and capacitance to ground at the output of your circuit as it exists now, then use an emitter follower or opamp follower to drive a quite-low output impedance. This will make it independent of random loads to a large degree.

[GibsonGM]

Okay, how would one measure what that is haha? Bear with me I'm still pretty new at this and not particularly fluent in multimeter use yet.

Ask someone on here who is actually an engineer to do some measurements on your circuit...then send them some $$ when you sell it    
Generally, most/many builders go for determining whether it is "high" or "low", relatively speaking, and don't go all the way in to get an actual (accurate) number.

>> Just a quick friendly word - be careful if you're 'selling these to customers' but don't have the background to explain how something works, how to obtain a measurement, etc.   I see you're learning, so keep it up, it will help you with the selling part, too!!  You sure don't want to have an issue because you did something you didn't understand and it caused a problem somewhere...happy building!

[Otbaudio]

Thanks for the tips guys, I'll give some of those a go in the morning and see what happens!

[Rob Strand]

Just measure the voltage with no load (VNL) with the load (VL), then output impedance is,

         Zo  = Zi (VNL/VL  - 1)

or      Zo  = Zi (VNL - VL) / VL

Any meter inaccuracies scale out due to the ratiometeric measurement.  (Meter offsets don't get removed but it's usually better if you don't remove zero offsets for multimeters on AC ranges.)

If VL is too close to VNL you might not be be able to get an accurate measurement.  Try to make VL drop to at least 0.9*VNL, maybe 0.5*VNL.   If Zo is very low you need a lot of meter digits to get an accurate measurement.

http://static.asset.aparat.com/avt/6901194-3454-b__820783082.jpg

[PRR]

> sell these on my Reverb store, and just trying to get the specs sorted for customers.

Make up a number. If you can't measure, neither can your customers.

Ford cars, same engine, rated 195HP or 200HP. They just made-up numbers. Later the numbers were supposed to be honest, but was my T-bird 137HP or 143HP? (Both numbers appear in specs; and there may be an explanation.) HOW do I know my Honda is 148HP? Even a chassis dyno test is costly and will read low (gear friction). No customer is gonna pull the engine and test it properly.

I don't see a need for a 'scope. You want a relative change of amplitude and quality. Clip-lead 1Meg, 100K, 10K, 1K resistors across the output. Play through it. At some point it will sound softer, muffled/thin, or distorted. If 1Meg is very-small change and 100K is a big change, try 330K. You really do not need to be more exact than that for rock-and-roll (or polka or tekno-salsa or....).

The output impedance can (usually will) be different for different frequencies. Try this with a naked turned-up guitar, 10K is a small change to deep bass but a big hit against the high treble. Strictly you want the value which gives "any" significant change.

A 220K output resistor works with the output cable to make a high-cut filter. 220K is generally awful high even for guitar-cord work (intolerable in any other audio). 10 feet of cable is 300pFd. 300pFd against 220K is 2.5KHz, near the top of what a guitar pickup will deliver. So try your "magic 220K" with a 1-foot cable and a 30-foot cable. You may find that it is much too sensitive to output cable (and other loading as R.G. says). You might really want 22K and a 3nFd cap, so cables from 1 to 50 feet sound all the same.

220K around AC-power wires begs for hum/buzz pickup.

[Otbaudio]

Thanks for your help on this guys, put some of your suggestions to use this morning, managed to get the output impedence down to 200Ω with no harsh topend! Sounds even better than it did before in fact! So massive improvement

[ElectricDruid]

Excellent work. That 220K series resistor was scaring me, for one. 220R sounds much more like it.

If it's an op-amp output and you've got 220R in series, you could reasonably quote "<1K output impedance" and no-one will care. That's low enough not to matter, and big enough you can't be caught out in the unlikely event that someone did actually measure it.

Tom

[Otbaudio]

Yeah good idea, thanks Tom. It's actually a Jfet output buffer I've added after what was the output capacitor and then the 220k, so the tone filtering is all the same, then just through an additional decoupling capacitor +the 200Ω resistor after the jfet which is mainly there to limit any switch pop.

Whilst I'm at it how do you calculate the current draw? And if it's done by measurement which parts of the circuit do you measure?

Tone I can do, specs I cannot haha

Oli

[ElectricDruid]

Current draw is simple - just put your meter on the "milliamps" setting (hopefully it won't be more than that!) and put it in series with the positive power lead. Not every unit will be identical, so give yourself some margin of error, or test a few. Depending on the components you're using, they should be close. Also remember to test the current with various control settings. The amount of current drawn might not be the same at all positions.

When I worked doing QA for an electronics company, the current drawn was a useful guide to whether a unit was working. The majority of faults would either cause a unit to draw too much current or too little, so it was a quick test to find faulty units and the given current went down on the test sheet associated with each unit. Of course, you can't do that alone, but it's a good test. If you're building to sell, I suggest having a series of basic tests like that and marking down current drawn in a spreadsheet for each serial number. If a unit comes back to you for repair, you'll be able to check whether something has changed significantly.

Sorry, wandered off a bit there.

T.

[antonis]

It's actually a Jfet output buffer I've added after what was the output capacitor and then the 220k, so the tone filtering is all the same, then just through an additional decoupling capacitor +the 200Ω resistor after the jfet which is mainly there to limit any switch pop.

Can you plz post a schematic..??

Sorry for the inconvenience but I can't imagine a de-coupling capacitor on a buffer output..

[Rixen]

plenty of smartphone signal generator apps available. Try Keuwlsofts Dual Channel Function generator app, sine, triangle, square, white noise, pink noise, modulation and sweep. (Android).