Gauge PM-1073 pre amp - lose high frequencies / schematics needed

Hi
I have a Gauge PM-1073 pre amp, It's a 1073 clone.
Lately I noticed that there's a lose of about 3 -10 db on frequencies above 3 khz comparing to other pre amps I have. (see the attached picture a comparison of 20 hz to 20 khz sweep on 3 different pre amps taht I have).

Since I don't have the schematics I can't really fix it.
What would be the suspected components that I should check first?
Here's few pictures of the PCB:





Thanks,
Asaf

Never heard of it but it looks like Neve 1073 preamp with an output level pot between the gain stages and output stage so you could look at schematics online and interpret accordingly.

Does the problem occur when the potentiometer is in the middle position (meaning at 12'oclock)?

This is a 1073 minus the equaliser. As such it is a 1290 in Neve terms...
1073 and 1290  schematics are widely available.

From what I can see from the pictures you posted this is a variant of the various chinese Neve copies. It is not that far from a Golden Age GA-73 or the like.

The problem you encounter is most likely caused by the (crappy) iron used in this clone.
My recommendation is to replace them all by quality iron (Carnhills or even used Marinairs (pricey)).
BTW: Neve used 22uF tantals instead of the 22uF electrolytics. Cloners tend to ignore this although they contribute to the sound...

chefducuisine said:

The problem you encounter is most likely caused by the (crappy) iron used in this clone.

Click to expand...

Really? A -20dB loss at 16kHz? Someone would have to go out of their way to make a transformer that crappy.

I have noticed some weirdness when using a level pot into the output stage of this circuit which is why I asked if the behavior persists with the pot in the 12 o'clock position.

I have seen really weird transformers in Chinese clones. Just have a closer look to the output transformer in the pictures above. To me this looks more like a cheap power transformer with way to thick wires than a proper LO1166 copy. The ones I had on the bench measured horribly. No one took care of adjusting the Zobel network accordingly and this can indeed cause these weird frequency plots. But I agree -20dB @ 16kHz is pretty heavy...

If I remember correctly the 'Gauge PA-1073' appeared around 2014 as an alternative to 'GoldenAge' but there were quality issues.

It would be helpful if the topic starter would answer a few questions (yours included):
- was this behaviour always as it is or did the preamp suddenly start sounding dark?
- how was the amp connected for the measurement (of high interest is the output loading here)?
- how was the gain set during measurement (Gain switch and output pot position)?
- was the line in or the mic in measured?
- was any built in input loading switched in circuit  (low Z)?
- was is the ohmic value of the output pot (it should be 4k7 or 5k)?
- what are the output transformers Zobel network values (the two parts near the yellow and green wires of the output tranny; I bet these are 1k5 and 10nF)

@smilan: It's your turn. Help us to help you.

Thanks Chefducuisine

chefducuisine said:

- was this behaviour always as it is or did the preamp suddenly start sounding dark?

Click to expand...

I got is second hand from Ebay so I don't know how it used to be before...

chefducuisine said:

- how was the amp connected for the measurement (of high interest is the output loading here)?

Click to expand...

I connected the output jack to my motu audio interface and also connected a scope to the output jack.

chefducuisine said:

- how was the gain set during measurement (Gain switch and output pot position)?
I've tested it in many different positions ant it acts same way

chefducuisine said:

- was the line in or the mic in measured?

Click to expand...

I've checked them bouth

chefducuisine said:

- was any built in input loading switched in circuit  (low Z)?

Click to expand...

There's a 300 Ohm / 1200 Ohm switch and it acts same way on both Zs

chefducuisine said:

- was is the ohmic value of the output pot (it should be 4k7 or 5k)?

Click to expand...

5K

chefducuisine said:

- what are the output transformers Zobel network values (the two parts near the yellow and green wires of the output tranny; I bet these are 1k5 and 10nF)

Click to expand...

Yes 1k5 and 10nF

@smilan: It's your turn. Help us to help you.

Click to expand...

I made some more testing with the scope and I found where the high frequencies drop starting (on the side of R1 that facing to TR1).


I replaced R1 and took off TR1 and still there's major drop between 1khz to 16khz.
Which components on this circuit can make this drop on high frequencies?
(C2 / C3?)

Assuming your tests and schematic are accurate [1], it would suggest that perhaps TR1 is at fault because the low pass filter of 2k2/4.7n is 15kHz but normally with TR1 in place, there would be a bootstrapping effect that makes that capacitor look smaller at AC frequencies. So by removing that transistor you definitely have broken it now because now you have the 2k2 with 4.7n to connection K which is connected to 1k5 and 100u to ground. So now it's quite expected to see a rolloff at 15kHz. So you can't do a meaninful measurement without TR1 in there. With the TR1 in, the follower action of the transistor will increase the AC impedance of that capacitor (C3) making it look smaller and thus pushing the rolloff above 20Khz.

So perhaps you can test that transistor or find a replacement. You could put just about any transistor with the same pinout and vaguely decent gain (like the ultra common general purpose NPN 2N3904 but pins reversed) although it may have a small negative impact on noise performance. If the high frequencies are restored the high frequencies in which case you have confirmation that that is the issue.

[1] How did you measure the frequency response at a point that has DC? If you probe does not block DC or if it's impedance is low or if it's capacitance is high, your measurements may not work in which case you're chasing your tail. Describe your probe and test apparatus. You really need to use a proper scope with a proper probe and compare a 1kHz test tone to 20kHz test tone. But if you make a high Z / low C probe and feed that into your sound card, that can work too if you know what you're doing.

squarewave said:

Assuming your tests and schematic are accurate [1], it would suggest that perhaps TR1 is at fault because the low pass filter of 2k2/4.7n is 15kHz but normally with TR1 in place, there would be a bootstrapping effect that makes that capacitor look smaller at AC frequencies. So by removing that transistor you definitely have broken it now because now you have the 2k2 with 4.7n to connection K which is connected to 1k5 and 100u to ground. So now it's quite expected to see a rolloff at 15kHz. So you can't do a meaninful measurement without TR1 in there. With the TR1 in, the follower action of the transistor will increase the AC impedance of that capacitor (C3) making it look smaller and thus pushing the rolloff above 20Khz.
So perhaps you can test that transistor or find a replacement. You could put just about any transistor with the same pinout and vaguely decent gain (like the ultra common general purpose NPN 2N3904 but pins reversed) although it may have a small negative impact on noise performance. If the high frequencies are restored the high frequencies in which case you have confirmation that that is the issue.

Click to expand...

While I took TR1 out of the board I tested it with a multimeter (in diode test mode) and it seems to work right. I don't know if this test is enough but I can get a new NPN transistor tomorrow.

squarewave said:

[1] How did you measure the frequency response at a point that has DC? If you probe does not block DC or if it's impedance is low or if it's capacitance is high, your measurements may not work in which case you're chasing your tail. Describe your probe and test apparatus. You really need to use a proper scope with a proper probe and compare a 1kHz test tone to 20kHz test tone. But if you make a high Z / low C probe and feed that into your sound card, that can work too if you know what you're doing.

Click to expand...

This is my prob:

My scope have a DC mode so I connected the tip of the prob to R1 and the sleeve to ground.
Here's a little video I made with all components installed:

https://youtu.be/rLanCHaeGYs

So your test equiptment looks fine but are you doing this with TR1 out? If yes, then the roll-off is quite normal because you have 2k2 / 4.7n + 1k5 + 100u + gnd which is going to definitely lose HF. So you're probe is fine but you need to restore TR1. But you're definitely on the right track.

So R1 and some capacitance is making an RC low pass filter. Unfortunately, because of the discrete design of this circuit, the possible causes are not as simple as a resistor capacitor filter that you might see in a book. There are AC impedances with the transistor and it's effect on other components that must be considered. If I just run through each net one-by-one, some possible causes are:

C2: Perhaps this capacitor is larger than it's supposed to be and the Miller Effect is causing low frequency loss.

TR1: Perhaps this transistor is broken in a way that has manifested as very low gain or high capacitance.

C3: This could very well be ok but if what it is connected to is low impedance that would actually explain the 16kHz roll-off because, like I said before, 2k2 / 4.7n is 16kHz. This is why I suspect TR1.

Or, another more ominous possibility would be an issue with overall feedback. So R5 contributes negative feedback from the output transistors and transformer so that makes analysis a little more difficult.

Check the bias. There should be a larger 47R resistor near the power transistor in that circuit. Measure the voltage across it. What do you get? It should be ~3.3V (70mA * 47R). If you adjust that trimmer a little, does the roll-off change? Don't turn it up too high though or you could burn something out.

Mmm, I see the 47R in your pic looks a little burned ....

Thank you  Squarewave,
I found a bad TIP122 transistor on the PSU that made problems with the bias and replaced it with a new one.
Now the difference between 1k to 16k on this board is reasonable and looks like it's  around the +/-0.5db difference that the manufacture defined.
So now there's only one problem, when I connecting the output jack to any external device the signal becoming distorted.
Here's the wave on the the  output when the preamp outpt jack is disconnected

And this is the wave on the on the otput when I connect the preamp output to an external device (I tried to connect it to line in on my console, directly to my sound card or to my zoon recorder)

This is a sign for a  a faulty transformer, or there's any thing else can make this distortion (problems with the zobel network?)?

What is the Vpp of that signal? What is the frequency?

All transformers distort at some level and more at low frequencies.

So this could actually be totally normal. That Neve output stage uses a specialty gapped transformer with DC running through it that is known to distort more easily. That is actually supposed to be what makes this circuit desirable. So if that signal is a 40Vpp 100Hz, I would expect it to distort. The only reservation I have is that the line in of something like a Zoom recorder is not going to load it much. If you load the output with 600 ohms I think it should distort more actually.

If you put <20Vpp and >1kHz into >1K load, I would think it should not be distorted at all. But I don't know the exact threshold at which it should start to distort.

I doubt there's anything wrong with that transformer. Transformers are not fragile parts that break easily. At least not in this scenario. You could probably completely short the secondary and nothing would happen. And a lot of cheap Chinese stuff is actually perfectly good. I would not be surprised if it was a spot-on match for the original Neve OT because, being a specialty gapped transformer for DC, they would have had to reverse engineer it and build it to spec or it wouldn't work at all. And they would save little or no money making it in a different way. So if it were wrong, that would mean they did something stupid and the Chinese are not known for being stupid.

squarewave said:

What is the Vpp of that signal? What is the frequency?

Click to expand...

I measured on the output 1Khz / 4Vpp
I'm sure it's not supposed to be like this...

Are you sure? Your picture shows 20V/div.

Also, where exactly did you measure? Note that you cannot measure the voltage across the primary of the transformer because of how transformers work. You have to measure the voltage across the load on the output.

Otherwise, no, it's not supposed to do that for sure.

Have you done anything with the bias trim? What voltage are you seeing across the big 47R resistor?

squarewave said:

Are you sure? Your picture shows 20V/div.

Also, where exactly did you measure? Note that you cannot measure the voltage across the primary of the transformer because of how transformers work. You have to measure the voltage across the load on the output.

Otherwise, no, it's not supposed to do that for sure.

Have you done anything with the bias trim? What voltage are you seeing across the big 47R resistor?

Click to expand...

My mistake, I confused the photos on my last post with some other measurements I took, this is the correct ones:

1Khz @ 4Vpp on the the  output when the preamp outpt jack is disconnected:


1Khz @ 4Vpp on the the  output when the preamp outpt jack is connected to the sound card:



I connected the probe tip to the tip of the TRS output jack and the probe's sleeve to ground.

I set the voltave on the large 47R resistor to 3.2V on one side and 0V on the side that connected to ground.

I'm not sure if you're going to get perfect results if you connect the probe tip to pins 2 or 3 of the XLR out and then ground to ground because the balanced output pins are not relative to ground - they are relative to each other (thus balanced).

Try it with a simple resistive load. Meaning just put a 1K resistor across pins 2 and 3 of the XLR output (without connecting it to the sound card) and put the scope tip on pin 2 and scope ground on pin 3.

However, if you connect to the sound card, pin 2 will likely be signal and pin 3 will likely be connected to the ground of the sound card in which case you should get good results with tip to pin 2 and ground to pin 3.

Although you might have an issue trying to put 4Vpp into a conventional computer sound card. It might be exceeding clipping diodes on the input. A cheap sound card might only be able to handle 2Vpp. The Neve can put out ~40Vpp so maybe you destroying your poor sound card. Does the sound card actually read the signal cleanly? Does the software have an oscilloscope mode? Is it clipping?

Does the distortion change with level? Meaning as you adjust the gain and level out does the shape of the distortion change?

If the problem isn't just a bad load (clipping diodes of sound card) or the way you're probing things then it might have to be something is still wrong with the output driver.

Finally I found the problem!
My sound card inputs connected thru a patchbay with the bus outputs of my console.
When I connected the output of the preamp to the sound card thru the front panel of the patcbay the resistance of the patcbay input was 3.2 Kohm and it seems like this  is what made the distortion.
When I connected the preamp output thru the insert return on my console the resistance was much higher, above the limits of my multimeter but now the signal is good and there's no distortion!
I've also tuned the bias in a way that she signal start to distort on both peaks simultaneously (now the voltage on the 47R resistor is 2.2V).

Thank you very very much Squarewave for the guidance and the support.

I've learned a lot from this repair.

Ok. But note that it should be able to drive loads as low as 600 ohms without distorting like that. Maybe your patchbay is wired in a strange way (like you're driving two outputs together). A proper test would be to load with 1K across pins 2 and 3 and scope across that as described in previous post.

I don't have in hand a 1K resistor so with 1.5K resistor between pins 2 and 3 on the output xlr I can drive up to 22.5Vpp with no distortion.

Sounds good. If you put two 1k5 in parallel and it does 20Vpp without visible distortion, it's solid and ready to rack.