# Output impedance measurement - without load?

#### Miniwatt

Hey everyone,

I would like to measure Zout of my amp. This requires measuring voltage at the speaker output with and without a load.
Now I've always been told that a working tube amp always needs a load. See my problem (?) here?
Would it be okay to have a dummyload at the output, measure the voltage and then briefly disconnect that dummyload? One second should be enough.
I'm not sure...

#### jan.didden

Paid Member
It's not necessary to have no-load and load. You only need two different loads to calculate, one does not have to be zero.
In fact, it is better not to use a zero load because zero load is a sort of special case and the calculation can be less accurate.

Jan

3 users

#### Miniwatt

Thanks Jan, my dummyload is two 50W resistors in series, I'll measure across both and then across one.

#### jan.didden

Paid Member
That's a perfectly good way.
When you say 'meausre across one', you probably mean: 'measure across the speaker output with 1 resistor load and with two resistor load'.
What are the loads in those cases?

Jan

#### Miniwatt

I was about to edit my last reply because I didn't make that clear.
Resistors are 4,7 Ohm, so the loads connected would be 4,7 and 9,4 Ohm.
Both on the 8 Ohm speaker output.
I hope I didn't forget anything this time.

#### TonyTecson

if you put a 100 ohm 10watt resistor on the 8 ohm tap, that will still be open circuit if you want to measure output impedance...

#### jan.didden

Paid Member
I thought so, but better to have it clear. That should work well.
Do you have the calculations available?

Jan

#### jan.didden

Paid Member
if you put a 100 ohm 10watt resistor on the 8 ohm tap, that will still be open circuit if you want to measure output impedance...
It is better to measure under normal operating conditions to have a better correspondance with the real use, so 4.7R and 9.4R are perfect.
A measurement with (almost) open circuit isn't very good.

Jan

#### Elvee

The multiple loads method is inaccurate and inexact for a number of reasons: inaccurate, because the calculations involve the subtraction of very close quantities, inexact because the impedance of an amplifier is never a pure resistance: it is partially non-linear, and it is complex (normally inductive).
A much better method is to drive the output of the AUT with a test amplifier, through the nominal load.
You then measure the residue appearing at the output of the AUT and compute the impedance using Ohm's law and the current through the load

3 users

#### TonyTecson

100 ohms is many many times bigger than 4 or 8 ohms to matter...

1 user

#### jan.didden

Paid Member
That external driving method has the same inaccuracies as you still have to deal with very small differences between the two situation.
But I see this is turning into a shouting match and bragging rather than helping out the OP, so I will leave.
The OP has my sympathy.

Jan

1 user

#### Diabolical Artificer

It is possible to power up a valve amp without a load, it's done sometimes to determine an amps LF stabilty. However it's done at a very low level. You can also use a capacitor as a load, again to test stability.

Andy.

1 user

Paid Member
1 user

#### Ultima Thule

Do also run the test with several frequencies as the transformer makes up a major bulk of complex set of apparent and stray impedance, say plotting the curve with a few frequency points starting with 20, 50, 100, 200.. etc Hz up to 20 kHz, in addition the test could be performed with different nominal power outputs, usually set at 1 kHz, say 1, 2, 5 and 10 Watts, or whatever may fit your amplifier, and redo the discrete frequencies sweep.
There you have some work to do, and, if and when done so, it would be nice to see your plots here, good luck.

1 user

#### Elvee

That external driving method has the same inaccuracies as you still have to deal with very small differences between the two situation.
No it hasn't: there is only one situation to deal with. If the amplifier has a very low Zout, the voltage can be small, but it is measured in isolation, not deducted from a much larger one.
The larger one (output of the test amplifier) is measured separately.
The method is simpler to implement than multiple loads, especially if the AUT is stereo: the other channel can be used as a driving amplifier

#### 45

I agree with @jan.didden. Even 10% relative error won't make any practical difference. The 2 loads method is good enough.

1 user

1 user

#### egellings

It's not necessary to have no-load and load. You only need two different loads to calculate, one does not have to be zero.
In fact, it is better not to use a zero load because zero load is a sort of special case and the calculation can be less accurate.

Jan
I will typically use an 8-ohm load for one measurement, then parallel that with additional resistance for the lower R load for the second measurement.

#### TonyTecson

It is better to measure under normal operating conditions to have a better correspondance with the real use, so 4.7R and 9.4R are perfect.
A measurement with (almost) open circuit isn't very good.

Jan
irrc, "Dogstar" built a high power tube amp with 100ohm resistor, permanently connected to the speaker output jacks, i suspect that the reason was to avoid the amp being played without a speaker connected, in my case i use 10ohm 3 watt resistor in series with a 0.01ufd cap to avoid such problems...