Pre-amps: input and output impedance

[gjm]

Is there such a thing as a rough guideline as to what is good, bad or appropriate?

A well-regarded pre-amp has the following:

Input Impedance: 100k ohms [this seems a fairly common figure]

Another that I've found details of has this specification:

Input Impedance: 25k ohms (50k ohms optional, 15k ohms minimum)

What difference would this make, or is it depdendent on the amplification circuitry and design of the individual amplifier? Output gain on both is similar; 10db gain in the first instance, and 8db in the second (although 20db is also available from the manufacturer).

I think there is some guideline here with regard to power amplifier compatibility, but I've no idea how that works.

[i_should_coco]

It depends what's driving it. Generally, the higher the input impedance and the lower the output impedance of whatever's driving it the better. Normally, at least a 1:10 ratio is desirable.

47kOhms is pretty standard. May valve amps are higher, while a good proportion of transistor amps are lower, e.g. 10kOhms.

[gjm]

So, the input impedance of the pre-amp should be at least 10x that of the output impedance of the device connecting to it... Does the same ratio apply to the connection twixt pre- and power-amplifier?

As a rough guideline, of course. I don't doubt there are exceptions.

[i_should_coco]

Yes. Pretty much anywhere there's such an interface.

[RichG]

Watch out if the input impedance is much below 10k as some weaker output stages may have trouble driving this low. I was experimenting with a power amp that had an input impedance of around 3k and found the sound was very flat and congested sounding even with a low output impedance op-amp based pre-amp. When changing to a different preamp which had better drive capability the sound improved enormously despite it having about twice the output impedance.
But in general the 10:1 rule works well.

[iaxel]

So if I'm trying to match a power amp (Accuphase) with a 20K input impedance and a pre-amp (Shindo) with a 5K output impedance - it would be a mismatch?
What kind of affect will this have on the sound?

[musicbox]

So if I'm trying to match a power amp (Accuphase) with a 20K input impedance and a pre-amp (Shindo) with a 5K output impedance - it would be a mismatch?
What kind of affect will this have on the sound?

That is probably going to roll off the treble and bass quite audibly, and flatten dynamics - it will probably sound very dull and boring in practice.

People always seem keen to recommend valve preamps with solid state power amps but they never seem to consider the output/input impedance question when they do so. The simple fact is that a lot of valve preamps just don't work well with a lot of solid state power amps.

[Valvebloke]

The first effect is that you will lose some signal. The voltage that the pre-amp delivers will be generated across the total impedance (20k+5k=25k). The voltage that the power amp 'sees' will only be the bit developed across its input impedance - 20k - so only 80% of the preamp's output voltage will appear at the power amp. Power is proportional to voltage squared. We don't care much about power at the pre-power interface. But the power amp is linear, so a reduction to 80% of voltage at the input appears unchanged at the power amp's output where it manifests itself as a reduction to 64% of the power you would otherwise have got. Of course if there is enough voltage headroom in the pre-amp (or sensitivity in the power amp) then you can fix this problem trivially by just turning the volume control on the pre up. You only suffer if the pre-amp is struggling to drive the power amp.

More subtle effects are not very important at these relatively low impedances, particularly if they are largely resistive (rather than inductive or capacitative). But for what it's worth a 5k output impedance means that the pre will find it a bit harder (strictly, slower) to charge up any stray capacitances at the interface - cable capacitance for example. For technical reasons the impedance we care about is 5k in parallel with 20k, which is 4k. If, say, you had 2.5nF of stray capacitance (that's a lot - a few tens of metres of interconnect cable) then the characteristic time for this to charge up through 4k would be 10 microseconds which is roughly the risetime of a 25kHz sine wave. So roll-off effects would start to become measurable up in the low ultrasonic region. I certainly wouldn't be able to hear them though and I don't think anyone else would in a blind test. Of course in the world where people can hear whether the fuse in the mains plug has had its terminals polished or whether the clamp holding the record down is made of the right kind of wood then these effects might be important.

Where you can get into more serious trouble is with pre-amps whose output impedances are rising into many tens of kilohms or where you're driving very low impedance loads (hundreds of metres of cable or lots of power amps in parallel). It is possible to build valve pre-amps that misbehave like this although there's no need to if you don't want to. The WAD Pre3 for example has such a low ouptut impedance that they say you can use it to drive an 8 ohm speaker and you can, I've tried it ! (I have to say it's not very good, mind.)

Valvebloke

[gjm]

Is there a simple way to measure output impedance of a pre-amp, or input impedance of a power amp? I'm thinking about numpties like me who have DIY power amps and have lost the build details...

[pure sound]

Put an appropriate pot across the output ie one that will barely load the pre. (47K or 100K) Put a sine wave through the pre. Measure the voltage developed across the pot, then adjust the pot until the voltage halves and measure the value of the pot at this position. That is your o/p impedance.

NB: output impedance can vary with frequency so this will give a value at the frequency chosen. In some instances the o/p impedance can rise at LF so it may be worth checking that it is tolerable.

[Valvebloke]

Depends what you mean by 'simple'. It's not trivial but whether you think it's tricky or not depends on your level of experience I guess. First, you need to make the measurement with an AC signal so it helps if you have a signal generator or a suitable computer sound card. You'll also need a reliable voltmeter.

To measure output impedance feed the signal into the pre's input and connect its output across a resistor, ideally one whose value, R1, is several times larger than the (suspected) output impedance. Measure the AC voltage across the resistor, call it V1. Then change the resistor for a lower value one, R2 - ideally something close to the output impedance. Without changing the input to the pre measure the voltage across the resistor again - V2. If I've done the sums right the pre's output impedance is then given by Rout = R1*R2*(V1-V2)/(V2*R1-V1*R2).

To measure the power amp's input impedance connect a resistor R3 in series with the input and apply an AC signal across the series combination of amp and resistor. Ideally R3 would be about the same value as the expected input impedance. Measure the voltage across the resistor - call this V3. Measure the voltage across the amp's input terminals - call this Vin. Rin = R3*Vin/V3.

I've just seen pure sound's recommendation. That uses the same physics as mine (one big resistance and one comparable with Rout) and should give a similar answer (mine includes the small correction for the fact that whatever the value of the pot, it will always load the pre a little bit). I agree with his qualifications about frequency.

VB

[iaxel]

The first effect is that you will lose some signal. The voltage that the pre-amp delivers will be generated across the total impedance (20k+5k=25k). The voltage that the power amp 'sees' will only be the bit developed across its input impedance - 20k - so only 80% of the preamp's output voltage will appear at the power amp. Power is proportional to voltage squared. We don't care much about power at the pre-power interface. But the power amp is linear, so a reduction to 80% of voltage at the input appears unchanged at the power amp's output where it manifests itself as a reduction to 64% of the power you would otherwise have got. Of course if there is enough voltage headroom in the pre-amp (or sensitivity in the power amp) then you can fix this problem trivially by just turning the volume control on the pre up. You only suffer if the pre-amp is struggling to drive the power amp.

Thanks for the detailed response!
So if, for example, my preamp has a measured output of 5V and the input sensitivity of the power amp is 0.6V - am I good to go? or should I look elsewhere?

[Valvebloke]

Sounds like you're good to go. In fact you might have a little too much coming out of the pre. (Just to be sure, are both these numbers RMS volts and are they both at full output ?) The only problem with this is that you may find you're only using a small part of the volume control's travel, right at the bottom. This could be fixed fairly easily with an inline attenuator between the pre and the power.

Incidentally I just had a quick look at a Shindo circuit and I see that in that pre at least he used a 0.47uF output capacitor. With just a 25k total circuit impedance the RC time would then be 12ms so there would be some low-frequency rolloff. It would be possible to design the inline attenuator to effectively raise the input impedance of the amp+attenuator quite a bit. That could lead to an extension of the system's bass capability.

VB

[iaxel]

(Just to be sure, are both these numbers RMS volts and are they both at full output ?)

Well, I really don't know. here's the spec sheet from their website:
http://www.shindo-laboratory.co.jp/E...urieges-L.html

[musicbox]

So the rule-of-thumb says you want at least 10:1 ratio between in/out impedences on power/pre...... in your case you will have a 4:1 ratio between the Accuphase and the Shindo. Can you demo these before you spend you money? If not.....

[griffy]

This has confused me a little, My Densen B200 pre-amp does not list its impedance but it does state that it can kick out upto 6 watts on its own.

Putting this into a power amp with say an input impedance of 50k would mean I'd have very little movement of the huge Densen volume pot to get it sounding very loud?

I'm guessing either this would not be a good match, or that I'd need some in-line attenuators?

[SergeAuckland]

This has confused me a little, My Densen B200 pre-amp does not list its impedance but it does state that it can kick out upto 6 watts on its own.

Putting this into a pre-amp with say an input impedance of 50k would mean I'd have very little movement of the huge Densen volume pot to get it sounding very loud?

I'm guessing either this would not be a good match, or that I'd need some in-line attenuators?

Densen's statement is meaningless unless the impedance that it can generate 6 watts into is specified. If we assume that this is 8 ohms, then 6 watts into 8 ohms means a voltage of 6.9 volts, which is perfectly reasonable for a pre-amp, a little high, but nothing out of the ordinary. It means that the Densen will have a low output impedance and it can drive pretty much any length of cable.

S.

[griffy]

Thanks Serge. Sounds like I have nothing to worry about then!

Densen do state that the reason for this is so it can drive long cable runs

From the volume potentiometer the signal is sent to the final stages of the B-200. The last stages are designed to be able to drive long cable runs, enabling it to work perfectly with several power amplifiers at once, even if they are placed alongway away from the B-200. This is done by making the output-stage a 6W Class-A power amplifier with no feedback

[nat8808]

Would there be a problem with an excessively high ratio?

The valve amps I intend to set up soon (Nestorovic) have an input of 220Kohms - they have two x RCA inputs for a possible balanced input so I presume 220K is with both sockets.

My current balanced pre appears to have an output of 600 ohms .. so a ratio of some 350:1 roughly. Is that going to cause any problems?

Might be that the 600ohms is a studio standard or something for balanced and mixing studio with hifi isn't always going to work...

[SergeAuckland]

Would there be a problem with an excessively high ratio?

The valve amps I intend to set up soon (Nestorovic) have an input of 220Kohms - they have two x RCA inputs for a possible balanced input so I presume 220K is with both sockets.

My current balanced pre appears to have an output of 600 ohms .. so a ratio of some 350:1 roughly. Is that going to cause any problems?

Might be that the 600ohms is a studio standard or something for balanced and mixing studio with hifi isn't always going to work...

600 ohms used to be the studio standard, and is still used when driving very long analogue lines, although long analogue lines tend not to be used much these days. If the line is long, then the signal gets digitised and multiplexed into something like MADI.

For home HiFi, a 600 ohms output impedance is a bit high, but perfectly OK, especially if the pre-amp has a valved output. A 600 ohm output normally feeds a >10k bridging input.

S.