Germanium preamp output impedance

Hi Folks,

I'm not too familiar with germanium transistor circuits, so I have a question about the output impedance of this one...see attached schematic.  It is a three transistor circuit with input transformer - only the relevant output stage is shown.  Basically, as it sits it will drive a 100k load just fine with 1% THD+N at +7dBm output.  Change that load to 600 ohms and the output level drops like a rock, -15dBm or so.  Next I tried a 15K:600R transformer on the output, but the max output I got was around -5dBm.

So my questions are:

A) I thought transistors had a low output impedance.  Is this not true of early germaniums?  Or is the 8k2 resistor responsible for this?

B) If an output transformer won't work, can I make it impedance balanced?  Would that require an 8k2 resistor or does the cap change things?

> I thought transistors had a low output impedance.

No. Common-emiter stages have a nearly infinite output Z. Resistor-loaded common-emiter stages have output Z essentially the collector resistor.

It "can" be lower if there is negative voltage feedback. Yours has none of that. It does have negative current feedback, which INcreases output Z, but not enuff to notice here.

Impedance made low with NFB does NOT change the power relationships. You have a 5K resistor in the path from power to output; it isn't going to get anywhere near 12Vpp out into anything like or lower than 5K load.

Plus that 8K2 on top of the ~~5K1 collector resistor, so ~~13K Zout.

> will drive a 100k load ... +7dBm output.  Change that load to 600 ohms and the output level drops like a rock, -15dBm or so.

+7dBu (not dBm) is 1.7V.

-15dBm is 0.13V.

About a 12:1 drop of output.

Counting on thumbs, we suspect Zout is like 12 times 600 Ohms, 7200 Ohms.

This does not agree. The 1.7V into a load almost 10X higher than Zout makes sense. Ass-uming Zout really is 13K, we expect 600r output to be 600/(13K+600) or 0.044 times as large, 0.075V, -20dBm. However I would believe this 12V scheme could manage 2.8V output (8V p-p) and then 0.13V in 600r is believable.

> I tried a 15K:600R transformer

A reasonable match; but what you gain with less preamp loading is largely lost with transformer step-down.

A hasty estimate of bias says 1.2mA, collector sitting at ~~6V. DC power into the stage is 14.4mW. The very-best a resistor-coupled "power" amp can do is 8% efficiency. So the maximum output possible is 0.08*14.4mW= 1.15 milliWatts. Near-enough zero dBm (+0.6dBm). That's peak at clipping. (I am ignoring the 8K2 resistor.) Reasonable headroom is much more than 10dB though maybe not 20dB. Your program level out must be -15dBm or less or it will mangle the program.

Why do you "have" to drive 600 Ohms? No modern gear needs that. This 5K-13K output is a reasonable interface to a single 10K-22K input typical on modern gear.

Why do you need even +7dBu? Most modern inputs can be potted-up to accept much lower levels and bang their meters.

The output impedance is defined by R4+R3, same as if they were Si transistors. Is not absolutely exactly that but take that number, the other effects are not a thing to consider in this case. So you have 13k3 output impedance and only 12V PS so you can't ask much power from it. Also Ge transistors are not good for anything which needs some power, they start to heat up and cause all sort of side effects you don't want. (change in gain, current, bias, etc)

EDIT... this first part is with rough numbers, lower I guess the 7dBm were in fact 7dBu into 10k and made better numbers with that, for the expected distortion.
You could take R3 out and get a lower impedance but that's not good if you have a short circuit at the output making the transistor to deliver more power than designed to. Best case you will be driving 600Ω from a 5k1 source, you will get about 1Vpp. Ignoring that, you have an output stage (without R3) capable of .003W (12V - 1V from the diode and transistor BE, 11V divided by 2 divided by sqrt(2), about 4V rms output best case, squared and decided by 5k1Ω) which are equivalent to about 4.7dBm.

If you want to drive a 600Ω load and have a perfect matching transformer without insertion loss you will get 4.7dBu output. There is no way this thing will drive +7dBm nothing, maybe +7dBu at 10k but that's not even close to 7dBm, which are the same as dBu when using a 600Ω load, for +7dBm into 10k you need 20Vpp signal, without taking into account the output impedance, so in the load.

EDIT... here starts the new numbers:
I haven't taken into account the distortion, if you were referring to 7dBu into 10k load that's about 1.73V rms or 4.9Vpp into 10k plus something loss on the 8k2 resistor, so about 3V into 18k load. That's about -3dBm, so there is no way to have more than -3dBu or dBm into a 600Ω with this output stage and keeping that 1% distortion. For that you need a 5k:600 transformer and take out the 8k2 resistor. Adding the insertion loss of the transformer you won't get quite to -3dBu but slightly less.

Maybe PRR already put something of this but I post anyway... Good luck!
JS

.

PRR said:

.

Click to expand...

2N2222 is germanium?? Good to know! 

JS

There is more 600 ohm input terminating equipment being offered for sale in 2015 than there was 15 years ago. 
LA-2a, LA-3a, 1176, Sta-Level, etc, from manufacturers and clones.

Most of them have really good imput sensitivity so you don't need to drive them to 4dBu or something like that to get good output level out of them, noise wouldn't be optimal but it wouldn't be in any case with germanium in the way. Most 1176 revisions are bridging input, if not all.

Still true a good thing to can drive 600Ω loads properly, you could have a separate stage for doing this just when it's needed, so you have two outputs, one "original" if this scheme is original to anything, and other with a silicon output stage as line driver to use when 600Ω loads needs it.

JS

mjrippe said:

Basically, as it sits it will drive a 100k load just fine with 1% THD+N at +7dBm output.  Change that load to 600 ohms and the output level drops like a rock, -15dBm or so. 

Click to expand...

Are you positive about this? It looks like the first measurement was made with 10k load. PRR pointed at that.
Now I am convinced the operating point is not correct. The collector voltage should be about -8V, which is not correct for such a stage.
The correct operating point depends on the effective load, but you must remember that for peaks that go towards the rail, the drive capability is limited by the voltage divider between the colletor resistor and the load, whence for negative peaks, the transistor is generally capable of much more drive.
The consequence is that the idle collector voltage should be closer to the emitter than the rail.
By changing R5 from 13k to 22-27k, you should increase headroom by about 6dB.
That being said, you cannot expect much drive capability from a single 12V rail.

Do you have anything against adding an emitter follower stage to the output?  This will lower the output Z, and give you more available current.  I would also tent to agree with Abbey about the 12V rail limiting your output swing.  Can you go higher?

Hi Folks,

Thanks for all the great informative replies.  Let me clear up a couple points:

I am testing the unit with an Amber 5500 which offers termination of 100k or 600r and measures in dBv or dBm (or volts, watts, etc). 

I do not particularly care about driving 600 ohm loads, but would like something that is not so "Hi-Z" for driving modern studio gear.  Balanced would be nice, which is why I tried a transformer and asked about impedance balancing.

So if R3 is there for short circuit protection *and* for setting the output impedance, can it be lowered significantly without risking the transistor?  And am I just totally off base thinking I can use an impedance balanced out?

Maybe I will mess with R5 as Abbey suggested.  Not sure if I want to add the 2N2222 stage, but thank you PRR!  These are definitely going to be "flavor" or "color" preamps anyhow. 

I've attached the complete schematic if anyone cares to see it.  They are capable of about 80dB of gain from 150 ohm source into 100k load, but will definitely need a pad in front as the "sweet spot" is around -45dBm input!

Be careful when calling dBm as you pointed is power, it's dependent on the impedance, so if you are measuring with your 100k input load dBm is probably measuring dBu and in 600Ω load is truly dBm. I could be wrong but is my best guess.

Sorry I made the numbers with 10k load in mind, 100k is much higher than the output impedance so only the voltage from it could be taken into account, so no way of knowing how much power it can deliver to the load with the specified distortion, in which case you can't know the equivalent output for 600Ω. You could make the numbers only taking the internal 5k as load and use the voltage at the 100k load, I think is the best way of knowing.

I agree adding one more stage would be a good solution and you could add an internal controlled dummy load so you can emulate the load on the germanium output stage.

JS

joaquins said:

Be careful when calling dBm as you pointed is power, it's dependent on the impedance, so if you are measuring with your 100k input load dBm is probably measuring dBu and in 600Ω load is truly dBm. I could be wrong but is my best guess.

Click to expand...

Ahhh, I see your point there.  I will have to dig out the manual and find out if that is the case but you are probably correct.

> if R3 is there for short circuit protection

Maybe... but a simple shorted output will not hurt anything. I can picture (and have devised) an extreme case which *might* pop S3 (high amplitude square waves into a capacitive load), but it is unlikely; the amp may not be fast enough to bust itself.

I think R3 is part of a gain control or mix network out to the right of what you show. If this is a PA mixer/amplifier, a bunch of 8K R3s would come together on a mix bus, and be recovered by another stage a lot like S2 or S3.

i.e. this preamp is made to drive stuff inside the same box, not the outside world.

If you have the rest of the schematic you may find a Line Out stage to copy. If this is a 12V-only system it may end in a high-current stage with maybe 50:600 transformer to kick-up to Proper Line Level. If you don't have the parts, a LM386 and a 150:600 (or 4-winding "150:900") transformer gives a beefy output.

Thanks PRR.  This is the entire schematic (though I drew it myself so it may contain errors).  The preamp has screw terminals for input and RCA jacks for output.  Webster electric made reel-to-reel tape recorders for home use, so I would think that is the reason for the hi-z outs.  I will try bypassing R3 and see where that gets me.

OK, just did a quick test on a unit that has been recapped to see if it gives a lower THD+N.  I am giving the measurements in V or mV RMS to avoid the dBu/dBm confusion of earlier.

With R3 in circuit:
2.00V into 100K load at 1% THD+N
115.8mV into 600R load at 1%

With R3 bypassed:
2.02V into 100K at 1%
147.7mV into 600R at 1%

So perhaps I'll jumper out R3 and leave it at that.

mjrippe said:

OK, just did a quick test on a unit that has been recapped to see if it gives a lower THD+N.  I am giving the measurements in V or mV RMS to avoid the dBu/dBm confusion of earlier.

With R3 in circuit:
2.00V into 100K load at 1% THD+N
115.8mV into 600R load at 1%

With R3 bypassed:
2.02V into 100K at 1%
147.7mV into 600R at 1%

So perhaps I'll jumper out R3 and leave it at that.

Click to expand...

That is very strange.
Without R3: Z out=8800r  should be close to 5.1k nominal
With R3: Z=9800r the 8.2k resistance should be seen as a difference

Assuming these measurements have been done with the same input level, there should be a difference between the measurements into 100k of about 1dB. The actual difference of 0.08dB indicates that R3 is more like 1k.

Hi Abbey, actually I adjusted the gain control of the preamp to set the output to the point where it reached 1% THD+N.  I can run it again and not change anything except jumpering R3.