Measuring output impedance

Hi all…

I have read about the subject on various sites and unfortunately they come up with different answers… so, I thought I would like to give it a try here.

Lets say you have a apparatus and just would like to measure the output impedance… Is there an easy way to do this and more specifically how to hook it all up.

My guess is a given voltage from a signal generator and a measured resistor and a scope … should do?

Best regards

/John

Put a resistive load across the output and reduce the value of the resistor until the output drops 6 dB. Measure the resistor, it equals your output resistance.

johnheath said:

Hi all…

I have read about the subject on various sites and unfortunately they come up with different answers… so, I thought I would like to give it a try here.

Lets say you have a apparatus and just would like to measure the output impedance… Is there an easy way to do this and more specifically how to hook it all up.

My guess is a given voltage from a signal generator and a measured resistor and a scope … should do?

Best regards

/John

Click to expand...

Yes, basically inject an AC current  (sinewave through a known resistor) then look for the voltage developed. You can change the sine wave frequency to determine impedance wrt frequency.

JR

Impedance will be a function of frequency, so one number is not a 'right answer'.

I still have to rig up such a device, but the basic idea is that you drive the output of the device under test through a measured resistor, and then measure both the voltage across the resistor and the voltage across the device under test. The ratio of the DUT voltage to the resistor voltage multiplied by the impedance of the series resistor is the impedance of the DUT.

What I wanted to add to the jig were some very large, high quality coupling caps in order to be able to measure DC regulators accurately without worrying about AC coupling the generator. So, I have a pile of 10µF polypropylene caps, an aluminum box, and some connectors waiting to be wired up, but I've been too busy.

However, this jig and a nice test set (like an Audio Precision) that can do sweeps will let you generate a pretty accurate impedance sweep with relatively little hassle. I still have to finish my version, but that's how it would work.

Monte McGuire said:

Impedance will be a function of frequency, so one number is not a 'right answer'.

I still have to rig up such a device, but the basic idea is that you drive the output of the device under test through a measured resistor, and then measure both the voltage across the resistor and the voltage across the device under test. The ratio of the DUT voltage to the resistor voltage multiplied by the impedance of the series resistor is the impedance of the DUT.

What I wanted to add to the jig were some very large, high quality coupling caps in order to be able to measure DC regulators accurately without worrying about AC coupling the generator. So, I have a pile of 10µF polypropylene caps, an aluminum box, and some connectors waiting to be wired up, but I've been too busy.

However, this jig and a nice test set (like an Audio Precision) that can do sweeps will let you generate a pretty accurate impedance sweep with relatively little hassle. I still have to finish my version, but that's how it would work.

Click to expand...

I built the capability into my old TS-1 audio test set. There was a 51 ohm resistor in series with the sine wave output but before the dB meter take off... Sweeping the frequency range while looking at the dB drop gave an indication of impedance vs frequency. I included a look-up table in the owner's manual for  ohms vs. dB drop.

JR

> Put a resistive load across the output and reduce the value of the resistor until the output drops 6 dB.

Fine for resistive outputs.

Many modern outputs are NFB amplifiers and will clip all to hell for any sane signal level loaded-down 6dB.

An alternate technique is to verify it is fine no-load. Bring up a small but accurately readable level in no load. Then apply the nominal load. Do math.

A given output is known to be fine either no-load or 10K load. Bring it up to 100mV in no load. Slap 10K across it and measure the new output level. Say 97mV.

10K caused a drop of 3mV in 100mV. Or about 1 part in 33. Zout is therefore 33 times smaller than 10K. About 300 Ohms.

Again: you have a True 600 Output (which will typically have an inbuilt pad to "make it" 600r). Bring it to 100mV no-load. Slap 600r on it. Output drops to 50mV, the case you describe where Zload=Zout.

If you have to meet a Government Spec, you need to account for "no load" really being meter load, and the math gets messier. This is never needed in routine studio work.

This is valid only for the frequency tested at! Start with 1K, then spot-test 100Hz and 10KHz. Many cap-coupled opamp outputs will be <100r at 1KHz+ but rising past several hundred Ohms below 100Hz. Transformers will usually drop in bass and rise at the top of the audio band.

johnheath said:

My guess is a given voltage from a signal generator and a measured resistor and a scope … should do?

Click to expand...

That is the direct method, that's directly derived from the definition of impedance, and there is the indirect method that others have evoked, measuring the voltage drop. Both methods are indeed valid, but depending on teh DUT, one or the other should be preferred.

The indirect method is particularly unsuitable for power amps, that have a very low output Z, typically a few milliohms. Loading it in such a way that it produces a measurable voltage drop may probably run it into non-linear operation. For example an amp with 8 ohms nominal output Z and a Damping factor of 100 has an output Z of 0.08 ohm. In order to produce a voltage drop of 5%, the load should be 1.6 ohm. There is no guarantee that the amp works nominally with such a low load. And due to the many sources of error, the accuracy of measurement is bound to be worse than +/-20%.
That clearly advocates the direct method.

Regarding measurement of line outputs, it must be noted that some topologies react to the actual load they are presented. That's the case of many transformer-based output stages using negative-impedance drive or tertiary NFB. An obvious example of the necessity to load the amp is tube amps with significant global NFB, that most often start oscillating (or at least being unstable) when not loaded.  If using the direct method, it should be done with the nominal load connected, and some math to calculate the resulting value. As to the indirect method, I would use a comparison between the voltage under nominal load and half and twice load, and again do some math.

That being said, the indirect method is often much easier to implement.

Thank you all… I do not know really where to start asking here

So… to summarize it all?

The output impedance will vary with frequency… Just for curiosity at what frequency does companies measure the output impedance of their devices as a reference for the manual? Like…"our device has an output impedance of 80ohm" If I understand it correctly that number should be mentioned according to a frequency? Maybe it is?

There is a "direct" and an "indirect" way to measure… I really try to grasp the difference but for a tube line amp it is "better" with the direct method with a nominal load?

Ok and finally… this is my poor english or poor understanding of the "electrical language"… but you all seem to be on the same route when saying things like "place a resistor across the output". What I just need to clarify for myself is how this is done in reality.

Please correct me here if I am wrong: I keep the device running idle… just measuring at the output (meaning not feeding signal though the device) And if I hook up like the picture and measure the difference between A and B gives me a ratio in voltage drop which will give me the Out Z (if I understand PRR correctly)?

What values on R (resistor) and V (voltage AC sine) do you recommend?


Thanks again you all.


Best regards

/John

I think the subject is broader and more complex than you expect.

There is a difference between a nominal output termination impedance,  and output source impedance, and an output drive capability, which can be three completely different numbers.

Back when audio terminations dealt in maximizing power transfer a 600 ohm output would drive a 600 ohm input.

Modern interfaces use what's called bridging i.e source impedance is 1/10th or less the load impedance.

Since some around here deal with legacy gear that may use 600:600 terminations, these all need to be understood.

Finally some of these numbers are nominal not precise measured impedances. Back decades ago when I gave my equipment 600 ohm output capability that just meant it could drive 600 ohms without raising a sweat.  The larger practical difference between pro and consumer gear is that consumer gear often has higher source impedance that can cause HF loss in long cable runs (due to cable capacitance).

Confused yet? (sorry).

JR

> The output impedance will vary with frequency…

"May vary"?

Old large broadcast gear had a Resistor Pad on the output. This isolates the internal amplifier from the external loads. What the load "sees" is pretty-much resistors, not tubes transformers and caps. And resistors are (ideally!!) constant at all frequencies.

While all real resistors tend toward 100 Ohms by 1GHz, few-hundred Ohm resistors ARE their marked values from DC to far above the audio band. So we don't need to say what frequency.

When faced with a strange box, you do not know what is inside. The spec-sheet may be quite vague, or just wrong. You probably should check medium high and low frequency for similarity.

As I said above, a common modern-box output is an opamp (<1 Ohm) plus a 20-200 Ohm resistor, and sometimes with a coupling cap. Since this type output is capable of driveing >2K, and generally drives 10K, the cap may be as small as 10uFd. So the true Zout is say 100r down to 150Hz rising toward 1K at 20Hz. With the design >2K load this works alright; with 10K it works fine. But when feeding a low-Z *inductive* load (600r transformer) it can make bumps.

A very real question is: WHY do you want to measure? Nearly ALL modern audio assumes low-Z source to high-Z load. As long as they are quite different, the exact values do NOT matter (usually).

JohnRoberts said:

I think the subject is broader and more complex than you expect.

There is a difference between a nominal output termination impedance,  and output source impedance, and an output drive capability, which can be three completely different numbers.

Back when audio terminations dealt in maximizing power transfer a 600 ohm output would drive a 600 ohm input.

Modern interfaces use what's called bridging i.e source impedance is 1/10th or less the load impedance.

Since some around here deal with legacy gear that may use 600:600 terminations, these all need to be understood.

Finally some of these numbers are nominal not precise measured impedances. Back decades ago when I gave my equipment 600 ohm output capability that just meant it could drive 600 ohms without raising a sweat.  The larger practical difference between pro and consumer gear is that consumer gear often has higher source impedance that can cause HF loss in long cable runs (due to cable capacitance).

Confused yet? (sorry).

JR

Click to expand...

Thank you sir

No, what you write is understood… but since I build some stuff myself I thought that it would be interesting to see what the output impedance is… for the drive capability and so. Measuring that is something I would like to learn. I mean why not? The more I understand and can calculate the greater is the chance that I build good working stuff.

Best regards

/John

johnheath said:

Just for curiosity at what frequency does companies measure the output impedance of their devices as a reference for the manual?

Click to expand...

It is usually 1kHz, but I have seen it at 400 cps (sic), or more rarely at 1592 Hz; the reason for this seemingly strange value being that it's actually 10 000 radians/sec, and some impedance-meters were calibrated accordingly.

There is a "direct" and an "indirect" way to measure… I really try to grasp the difference but for a tube line amp it is "better" with the direct method with a nominal load?

Click to expand...

I would say for a very low output Z unit, the direct method is best, and for medium-Z units, which covers almost anything from consumer equipment to pro gear, and including tube and solid-state, both methods are capable of giving similar results. There are some cases of products with a high output Z, such as magnetic loop amplifiers and RMI industrial amplifiers where the indirect method is preferrable.

And if I hook up like the picture and measure the difference between A and B gives me a ratio in voltage drop which will give me the Out Z (if I understand PRR correctly)?

Click to expand...

On eof teh measurements give the voltage of your tone generator, the  other measures a voltage that results from the formula V2=Vgen.Z/Z+R+Rg
It is very important to take into account the source impedance of your generator.

What values on R (resistor) and V (voltage AC sine) do you recommend?

Click to expand...

Best accuracy is when R+Rg =Z and V should be about the nominal output level of the DUT.
That means if you have no idea of the nominal values, you'll have to use successive approximation.

PRR said:

A very real question is: WHY do you want to measure? Nearly ALL modern audio assumes low-Z source to high-Z load. As long as they are quite different, the exact values do NOT matter (usually).

Click to expand...

Thank you sir

Yes, that is a good question if I only had modern devices around… made by another person I guess.

The thing is, as I answered JR above, that I build some devices myself and would like to be able to calculate and if needed adjust for better working stuff. Basically I am curious and would like to know how it all works within electronics and more precise within audio. I know that is a vast subject but I take one step at the time… but I have no teacher or tutor so I read a lot but the information on the web can be misleading from time to time.

I know that my gear ends up with a quite low Z (100 - 200ohm) and they have a fine frequency response. But I have read about devices were an output transformer 10k:600R is used but the designer says it has an output impedance just above 50R. That made me wonder what my devices have. As I said… I am curious by nature

Best regards

/John

abbey road d enfer said:

On eof teh measurements give the voltage of your tone generator, the  other measures a voltage that results from the formula V2=Vgen.Z/Z+R+Rg
It is very important to take into account the source impedance of your generator.
Best accuracy is when R+Rg =Z and V should be about the nominal output level of the DUT.
That means if you have no idea of the nominal values, you'll have to use successive approximation.

Click to expand...

Thank you sir

Very nice and informative...

But still a bit confused about the formula. So voltage measured at "B" is what I get from the formula? And that equals the source impedance of the generator divided with the sum of the resistor value, the output impedance and Rg (boink!  )?

"Boink" means I hit my head in the doorway No, not really but where do I find that value?

Best regards

/John

johnheath said:

But still a bit confused about the formula. So voltage measured at "B" is what I get from the formula? And that equals the source impedance of the generator divided with the sum of the resistor value, the output impedance and Rg (boink!  )?

"Boink" means I hit my head in the doorway No, not really but where do I find that value?

Click to expand...

that should be in the specs, or you can measure it, using the basic indirect method - measure open-circuit voltage, then voltage under load.

abbey road d enfer said:

that should be in the specs, or you can measure it, using the basic indirect method - measure open-circuit voltage, then voltage under load.

Click to expand...

I am sorry but Rg is not understood for me here  :-\

Since I guess we are fiddling at the output I guess it could be the secondary of the output transformer, but all I know about Rg is it being the grid resistor… But I guess I am missing something valuable here

Best regards

/John

johnheath said:

Thank you sir

Yes, that is a good question if I only had modern devices around… made by another person I guess.

The thing is, as I answered JR above, that I build some devices myself and would like to be able to calculate and if needed adjust for better working stuff. Basically I am curious and would like to know how it all works within electronics and more precise within audio. I know that is a vast subject but I take one step at the time… but I have no teacher or tutor so I read a lot but the information on the web can be misleading from time to time.

I know that my gear ends up with a quite low Z (100 - 200ohm) and they have a fine frequency response. But I have read about devices were an output transformer 10k:600R is used but the designer says it has an output impedance just above 50R. That made me wonder what my devices have. As I said… I am curious by nature

Best regards

/John

Click to expand...

Good design practice is to define your source impedance by adding a resistor in series with a low impedance source (like an op amp output). The true output impedance of the op amp could change 10:1 over the audio band but still be only millOhms. Adding a 50-100 ohm resistor in series makes the milliOhm variation insignificant and adds the extra benefit of decoupling external capacitance that can destabilize negative feedback paths. 

Profession gear usually uses around 50 ohms for build outs (which is near the characteristic impedance of audio cables (or it was last time I checked decades ago). Consumer gear can use anywhere from a few hundred ohms to as high a 2k ohm.  :

JR

> transformer 10k:600R is used but the designer says it has an output impedance just above 50R.

Transformer ideally just levers V/I to show different impedance.

So 833r driving 10K:600r "should" be 50r.

BUT. Transformers have resistance. Typically 5%-10% of the nominal working impedance. So a "600r" winding may show 60r even when driven from a zero-Z source. So the "50r" may actually be correct for a good transformer driven direct by an opamp or other low-low source. (But we'd rarely want a 10K:600 in that path?) 

Some designers don't figure/measure right. Some designers have Marketing Men to "improve" their documentation into sales-sheets.

JohnRoberts said:

Profession gear usually uses around 50 ohms for build outs (which is near the characteristic impedance of audio cables (or it was last time I checked decades ago). Consumer gear can use anywhere from a few hundred ohms to as high a 2k ohm.  :

JR

Click to expand...

That is interesting to hear - thank you

Best regards

/John

PRR said:

Transformer ideally just levers V/I to show different impedance.

So 833r driving 10K:600r "should" be 50r.

Click to expand...

Thank you sir

Yes, so i have been told. But would that be possible in a tube preamp with the mentioned transformer ratio?

PRR said:

Some designers don't figure/measure right. Some designers have Marketing Men to "improve" their documentation into sales-sheets.

Click to expand...

Also true in many different matters and subjects.

Best regards

John