Measuring amp output impedance

[Chuck]

Is there an easy way to measure the output impedance or a tube amp when it is unknown?

[PaulC]

Yep - there's several.  The way I like to do it is to hook my variac across the primary leads (the ones that would go to the plates of the pwr tubes.  Then starting at zero vac I bring the voltage up while measuring the AC voltage on the secondary.  I tend to bring up the voltage to give me 1vac on the secondary (easy for math).  Then measure the voltage on the secondary, and this will give you the turns ratio.

So if it took 31.62vac on the primary to give 1vac on the secondary you would have a  31.62 to 1 turns ratio.

Then you sq the primary voltage to give the imp ratio:

sq31.62 = 999.82 to 1 imp ratio

Then times the number by what you think the speaker load should be:

999.82 x 4 ohm load = 3999.28

Then you have to know some tube imp numbers.  A pair of 6l6's run in a standard type of amp will  want about 4kohm which this looks like it will do.

Hope this helps!

PaulC
Heritage amps/Tim & timmy pedals

[Chuck]

Thanks Paul.

But, um, I think you missed the word "easy."   

Maybe I should just ask: How do you tell if the amp is matched correctly to the speakers.
Without opening the amp.

[GibsonGM]

That's easy, Chuck...neither the speakers nor the output transformer blows..
   
Are you building one, or doesn't yours say right on there what the load should be?

[Chuck]

To be honest, I was looking at the specs for the Ultimate Attenuator.
It says:

The basic unit can be used with 2, 4, 8 and 16 ohm amps or impedance taps and can be used with 4,8 or 16 ohm cabinets in any impedance combination except for a 2 ohm cabinet.

And I was wondering what it really was.  I know I'm using the 16 ohm output on a Marshall TSL100 and the 16 ohm input to a 1960a cabinet.  And I know I've driven the hell out of the UA, with the Marshall cranked and the UA turned down to bedroom levels -- with no smoke or flames.

But I'm considering getting a new amp and don't want to smoke several grand away.  Since I was having trouble understanding how the impedance range can be so great, I thought I'd measure to see what it really is.

[roknjohn]

Thanks Paul.

I have an old Supro amp without a speaker I've been wanting to fix up for some time. I was always wondering how to go about determining (with some accuracy) what the original speaker might have been.

I don't have access to a variac, but I assume you could do the same using an audio signal generator, you would just end up with smaller voltages and probably not the nice round numbers as you descibed.

[Meanderthal]

 That old Supro- a combo? If it's a Corsican, I can check my Harp player's amp... all I know right now is it's Jensen...

The schematics might be available, and that'll show impedance...

[sfx1999]

Which tube amp are we talking about? Also, does it have a speaker?

[R.G.]

There are two impedances you might be interested in.

1. The intended speaker impedance to be connected - that is, 4 ohm, 8 ohm, 16 ohm, etc.
2. The internal impedance of the amplifier itself. This is almost always much smaller than the speaker it drives, and is one measure of how tightly the amp can "grip" the speaker to control cone motion.

Which one did you mean?

[db]

I seem to remember from my study years ago about power transfer being most efficient when Rsource = Rload.  Obviously, this is a little more complicated when dealing with a typical speaker impedance.

[Chuck]

Since the thread got jacked and confused ......  I'll say it again.

I was looking at the specs for the Ultimate Attenuator.
It says:

The basic unit can be used with 2, 4, 8 and 16 ohm amps or impedance taps and can be used with 4,8 or 16 ohm cabinets in any impedance combination except for a 2 ohm cabinet.

And I was wondering what it really was.  I know I'm using the 16 ohm output on a Marshall TSL100 and the 16 ohm input to a 1960a cabinet.  And I know I've driven the hell out of the UA, with the Marshall cranked and the UA turned down to bedroom levels -- with no smoke or flames.

But I'm considering getting a new amp and don't want to smoke several grand away.  Since I was having trouble understanding how the impedance range can be so great, I thought I'd measure to see what it really is.

[db]

Well since you really want to be able to tell what the output impedance is without looking at the manual:

If you consider it as a black box producing an output voltage via a source resistance (simplified view),  the only way to deduce the output resistance is to measure the voltage output Vload (RMS with a test tone input) across a load resistance Rload and then vary Rload slightly either side of the supposed nominal source resistance keeping a note of both.

i.e. measure Vload1 for Rload1 from which you calculate Iload1 (=Vload1/Rload1)
and measure Vload2 for Rload2 from which you calculate Iload2 (=Vload2/Rload2)

Consider the internal voltage generator voltage as V in series with a n internal source resistance Rs:

V = Iload1(Rs+Rload1)
V = Iload2(Rs+Rload2)

so 0 = Iload1.Rs + Iload1.Rload1 - Iload2.Rs - Iload2.Rload2
so 0 = Iload1.Rs + Vload1 - Iload2.Rs - Vload2

from which you can calculate:
Rs = (Vload2 - Vload1) / (Iload1 - Iload2).

but this is theory and I wouldn't necessarily recommend that you try it unless you know what you are doing (and certainly not at full power output).  And in a real test, the load resistance would have to be a resistance of a high enough power rating.

[db]

Oh and also, if you really are spending a few grand on an amp, it may be wise to contact the manufacturer to ask their advice, as using a power soak may well void the warranty (I don't know this to be the case but better safe than sorry).

[roknjohn]

Chuck,

Sorry if I caused things to head off topic.

What are trying to measure the impedance of? The output of the amp? The Speaker cabinet? The Ultimate Attenuator?

Most (not all) amp heads have impedance that is selectable either by a switch or multiple jacks. These usually come from from different taps on the output transformer.

Speaker impedances vary with the speaker itself, as well as the speaker configuration within the cabinet.

If I'm being too basic, let me know. Perhaps you could tell us how you plan to use the new amp. Is it just a replacement for the Marshall?

It sounds to me like the Ultimate Attenuator will allow you to select a setting that will match your amp.

[roknjohn]

Sorry Chuck. JUst went back and read the original question again.

Sounds like you are trying to match the amplifier as closely as you can to either the speaker or the Ultimate Attenuator.

What specifically are your concerns about "Going up in smoke"?

Are you trying to match a new amplifier to your existing Marshall cabinet?

[brett]

Hi
How many people have had an amp "go up in smoke" due to impedance mis-matching?
As far as I can see, an impedance mis-match might cause some extra losses in the transformer, but power will be down as well. 
As mentioned above, maximum power output occurs with a match, not a mis-match:

I seem to remember from my study years ago about power transfer being most efficient when Rsource = Rload.

Here's an easy, but somewhat uncertain, method:  In many transformers, the ratio of DC:AC impedance is fairly consistent (often about 1:10).  You might check the DC impedances (and nominal AC impedance) of similar wattage output transformers (! only compare push-pull with push-pull and SE with SE ! aka class AB and class A).  I think you'll find that there's a fairly consistent ratio between transformers.  Hopefully, it will be consistent enough to work out what the AC impedance is when you measure the DC impedance of the secondary of an unknown output transformer with a DMM.   

Here's the principle behind the method:  Because there are some limits on keeping output transformers low cost and still efficient, there's a lower limit to the number of windings needed on the secondary, and an upper limit on the mass (and diameter) of copper wire.  These factors (and maybe others) result in a situation that's familiar in speakers: there's a somewhat consistent ratio of DC:AC impedance for a given power rating across impedances.  Ideally, we would want the ratio is as small as possible, because DC impedance doesn't create any sound from a speaker.  However, because copper wire in a speaker needs to be thin to accomodate lots of turns, some impedance is unavoidable.  From memory, a typical 8 ohm 50W speaker has about 2 to 2.5 ohms of DC impedance.  The numbers might be wrong, but I'm reasonably sure that the principle is right.  Output transformers should also have consistent impedance ratios (In my experience most power and coupling transformers seem to).  The only factors that could stuff this up are things like comparing a copper-wound and aluminium-wound transformers, or a really cheap one (less copper) with a really expensive one (more copper), or a toroidal (ring-shaped) one (rare) with a box shaped one (99%).

[PaulC]

Chuck,

   Sorry I didn't understand your question.  What it seems like you're asking is what is the input imp of the box, and what's it output imp - not how to measure amp output imp.

   That's a tricky question, and you'd be best served to contact the guy that makes the box. 

A lot of units that say they can take a wide range are set up to take an average load.  It'll get you close, but it wont nail it.  Ask the builder what's the optimum connection.

I don't have access to a variac, but I assume you could do the same using an audio signal generator, you would just end up with smaller voltages and probably not the nice round numbers as you described.

You can use a signal gen if it has a nice,low source imp.  The primary winding that you hook it to average out between about 80 ohms plate to plate to 300 ohms plate to plate.  I like to set it for 1vac on the secondary just to keep from having to figure out the ratio since you're shooting for "1" anyway.  If you can only get something like .5vac with a signal gen just divide the input voltage by the secondary voltage to get the ratio.

Later, PaulC
Heritage amps/Tim & timmy pedals

[d95err]

For solid state amps, the speaker impedance is important. Too low impedance, and the output transistors could blow because they will be supplying too much current. The "Ultimate Attenuator" is a solid state amp, so that's why it needs a 4 ohm or bigger speaker.

As far as the input into the attenuator is concerned - is there perhaps a switch to select the amp impedance?

[Harry Palms]

Is there an easy way to measure the output impedance or a tube amp when it is unknown?

Good luck.

Like db says, you Nortonize the circuit and calculate a load line. Each point down the load line will will represent an output voltage and current which your output dumps into any given amount of resistance. Yeah, you'll have a crapload of impedance variations; frequency dependant parameters constantly shifting the impedance values.

Not for the squeamish. Not for me, at least.

The odds are that it's a "low" impedance out. It's a current source after all, right?

The load is reactant, right? Let it determine it's own needs. The load line tells you this.

Too "low" of an impedance (or too high, even) will cause this missing the mark of "grip"

2. The internal impedance of the amplifier itself. This is almost always much smaller than the speaker it drives, and is one measure of how tightly the amp can "grip" the speaker to control cone motion.

transfer/waste of the energy you've worked so hard to refine.

The way it's wasted at the transformer?

Well, let RG 'splain that to you. He does this for a living.

[Holden80]

Hi, ran into this problem last night with a new amp, landed here looking for a solution.  Here's what I did:
Your amp head won't be harmed by a cab with too much resistance, but it'll be smoked by a cab that doesn't have enough. 
The highest impedance it's likely to want will be 16 ohms, and the options on the table will be 16, 8 or 4 ohms.  So, start with a 16 ohm cabinet.  If it sounds good, job done.  If it sounds muddy, or like you're listening underwater (speaking from experience), you'll want to try 8 ohms next.
Someone please let me know if there's something wrong with this logic, but it worked for me