Eros Phono finished! Plus question about gain

[dave-tx]

Ah yes, thanks for reminding me.  I haven't measured the preamp yet, but did some measurements on the Eros itself.  I think these results point more to a problem in my build than an issue with the preamp.

I hooked up a few test signals, here are the results.  I used a signal generator on an iPhone and measured with my scope probes.  Don't know if I needed to match any particular output load(?)

All measurements peak-to-peak, 20mV input

60Hz:  1.3V out
500Hz: 0.5V
1kHz:  0.35V
10kHz: 0.30V
20kHz: 0.22V

Both channels measured approximately the same.  It appears that even the 60Hz level is below nominal.  Or am I fooling myself because I'm not using an inverse-RIAA source (although 1kHz should be the 0dB point on the RIAA curve, yes?)?  Sorry for the clueless questions, I'm learning as I go.

Are there any good internal nodes to measure to try to track down the missing gain?

Many thanks for the continued help.
-David

 

[grufti]

The ratio between your 60Hz measurement and your 1kHz measurement is close enough at ~ 11.4dB. That difference is the result of the RIAA filter doing its job. This tells us that your meter is reasonably accurate at 1kHz, but it's still worthwhile to find out what type of meter you used for the measurements.

That being said, you should measure way more than 0.35V at 1kHz with 20mV in. The Eros should be right around 1.5V out with 5mV in. Something inside your Eros isn't working the way it should.

20mV is a bit much on the input for your measurements [5mv would have been better], but in this case it is definitely good enough to point to the culprit.

 

[dave-tx]

Good that the ratio between the frequencies seems right.  I was just using an old Techtronix oscilloscope to measure both the input and output signals, probes set to 1x.  So my readings involve some visual interpolation between grid marks, but they should be reasonably accurate.

I was initially using 5mV in, but 5mV/div is the smallest my scope will go, so wanted a little more resolution.  I will drop it down to 5mV input for any future measurements...that's closer to what my cartridge should be putting out anyway.

Thanks,
David

 

[grufti]

Your measurements at 10kHz and 20kHz are well below what they should be. 20kHz should be almost 20dB higher than the 1kHz value with the RIAA eq, but that issue is probably tied up with the underlying problem with your Eros. I would not worry about that now.

Your scope won't have any issues with measurement accuracy at frequencies higher than 60Hz. That's good to know.

PB will know where to look. He knows way more about the BH amps than I ever will.

 

[Doc B]

Are you using the same units for input and output measurements? Vrms?

 

[dave-tx]

Are you using the same units for input and output measurements? Vrms?

Pretty sure they are the same - It's probably 15 years since I've used my oscilloscope, but I believe both scope channels are set up identically.  My measurements should be peak-to-peak, since I'm "measuring" by visually looking at the waveforms.

Thanks,
David

 

[grufti]

You should poke around socket C in the middle and also check your 6922, E88CC, 7308 tube since both of your channels seem to have the same problem. Unfortunately the wiring is rather crowded right there. Check the continuity of your ground connection(s) in that area first (ground near socket C to input ground for example).

Check the tube itself and/or replace it.

Check all resistance measurements and voltages for the connections to socket C. It gets pretty crowded in there, but you can follow the wiring to the next terminal strip. You can use page 1 of the schematic to find your way around. The socket pins are shown in the schematic: C1 and C6 should be at bit less than 170V, C2 and C7 at about 100V, C4 and C5 are the heater, C3 and C8 are the cathodes. I don't remember off the top of my head what the cathode voltage should be. There are references to the corresponding terminal strip points in the schematic. What ultimately matters is what the tube sees at its pins.

Be careful measuring directly at the socket. There isn't much room for error. You probably want to connect your probes while the amp is turned off and then turn it on to measure.

 

[Paul Joppa]

20mV pk-pk is 7mVrms, a very good number for testing.

350mV/20mV (gain at 1kHz) is 25dB gain, well below the 48dB or so you should expect.

At 10kHz the gain is hardly down at all, it should be only 20% of that at 1kHz.

These data indicate an error in the RIAA equalization. I'm sorry but it will be a day or two before I can analyze it, hopefully either that's enough clue, or someone else will have the critical insight before then!

 

[grufti]

In this case the scope worked just fine to document that the gain from this Eros is not what it should be. When the difference between the expected result and the actual is as great as 24dB then reading peak-to-peak voltages from a scope will do the job. Quite a few RMS reading meters can't be trusted at 20kHz and just about any scope will still display 20kHz accurately ... trade-offs everywhere.

 

[Paul Joppa]

Sorry, no useful insight yet, beyond my previous observations. The RIAA equalization is not working correctly - there's too little bass boost, and hardly any treble cut - and the gain is more than 20dB too low at 1kHz. I see two further useful things to do:

* Experimentally, you can look with the scope at various intermediate points in the circuit. I'd suggest the EF86 grid (T19/T27), EF86 plate (A6/B6 or T4/T12), 6922 plate (T9/T10), and 6922 cathode (C3/C8). Use the scope AC coupled while you have 20mV peak to peak into the input. Your previous frequencies are fine for this.

* Mechanically, inspect the 0.030 and 0.010 capacitors to make sure they are the right values and installed in the right places, and measure the 10.5K resistor.

The resistances you measured earlier are consistent with the circuit diagram, while the ones in the manual are not. Being away from my own Eros, I can't confirm either set yet, but I thought I should mention that. Those resistances indicate to me that the resistors in the RIAA, and the plate and cathode resistors, are all the correct value, hence my emphasis on the capacitors. Your DC voltages indicate the correct DC operating points.

A couple faint possibilities also occurred to me: (These are faint because I imagine you have inspected this things way too many times already - but I list them because they are possible causes of the observed symptoms.)

* Check the integrity of the grid and plate stoppers, attached to A9, B9, C2, and C7 (220 ohm) and C1 and C6 (100 ohm)

* Check correct attachment and good solder joints and correct orientation of the 100uF/160v electrolytic capacitors - I can see the possibility that without this capacitor, negative feedback to the screen grid of the EF86 would reduce gain. See p.28 of the manual.

This one is a real puzzler! At least we now know for sure that it is inside the Eros.

 

[dave-tx]

Thanks to all for the suggestions.  I'll get the scope on the intermediate nodes tomorrow evening and report back. 

I double-checked the 0.030 and 0.010 caps and the 10.5k resistor, all are in the correct location and solder joints appear to be solid. 

Since the identical lack of gain is present in both channels, does that tend to point to a common component to both channels (assuming, for now, that it's not a swapped component in both channels)?  Is it possible that the 6922 isn't functioning properly?  I have no experience at all with the behavior of a misbehaving tube (would that even manifest in both channels?) and have no idea how to test the tube itself, other than visually verifying that the heater glows.

Grid and plate stoppers all appear attached well and measure to the correct resistances, and the 100uF caps are oriented as per the manual.

Would there be any value in trying to bypass the RIAA equalization to identify it or eliminate it from the equation?

Thanks for the help, everyone.  I'm glad this is a puzzler (so far), thinking about it is definitely a learning experience!

-David

 

[dave-tx]

Quick update with measurements at the intermediate points mentioned by Paul.

Measurements were approximately the same for both channels.  Measured visually, peak-to-peak, with an o'scope.  20mV input sine wave.

Measurements are at 60Hz, 500Hz, 1kHz, 10kHz, 20kHz

T19/T27: 20mV for all frequencies
T4/T12: 600mV, 220mV, 200mV, 50mV, 30mV
T9/T10: 800mV, 300mV, 200mV, 200mV, 150mV
C3/C8: 600mV, 220mV, 200mV, 50mV, 40mV

 

[grufti]

I can't make any sense of the values that you measured without digging into the schematic and I won't be able to do that before Saturday. I hope you get help from either one of the Paul's much sooner. Something is fishy, but what? I bet you didn't need me to tell you that.

In response to Grainger: what really matters at this point are the ratios between the voltages. As long as all values are either RMS or peak-to-peak everything is pretty straightforward.

 

[dave-tx]

Is there any value in disconnecting the EF86 plate from the RIAA filter, leaving only the 75kohm load resistor and checking the gain at the plate there?  Or will it be too hard to know what the correct gain should be without the RIAA filter in place?

Similarly, is there any way to isolate the 2nd stage 6922 to test it on its own?

Thanks,
David

 

[Paul Joppa]

Bingo!

The 100uF is not connected. Perhaps it is in the wrong holes? - there are several in a row.

I determined this because there are large AC voltages on the 6922 cathodes (C3/C8), which should be bypassed to ground by that capacitor.

With respect to questions about the tubes, at least in this circuit, the fact that the DC voltages are right indicates that their gain parameters (transconductance and/or mu) are within specs.

 

[dave-tx]

Thanks, Paul!  I won't be able to check until late tonight (or early Saturday morning, depending on how many beers the evening brings), but I'm betting you're right about the wrong hole. 

Will update...

-David

 

[Grainger49]

Dave,

You are a true Texan.  I remember fond nights at Up In Smoke, a BBQ joint North of Fort Worth.  Great place, great wings and very fresh Lone Star beer!

 

[dave-tx]

Not sure if my neighbors would consider me a true Texan (I'm originally from Wilmington/Philadelphia), but I've been here long enough to love good BBQ!

Took a quick look at Page 29 in the manual, and although I have the 100uF cap in the right place, I am missing the 1" red wire that connects the cap to breg...on both boards, of course.  Essentially, as Paul stated, the cap is unconnected.

Will add this in the morning and report back. 

Thanks,
David

 

[dave-tx]

Updated output voltages...All look good relative to each other, particularly considering margin of error on visual scope measurements.

60Hz: 40v
500Hz: 11v
1kHz: 7v
10kHz: 1.7v
20kHz:1v

And of course, the only test that really matters...Put it back in the system and delight ensued.  I have a feeling I won't be getting much done this weekend other than listening.  And I'm okay with that.

Thank you very much, everybody, for patience and help in finding this mistake.
-David

 

[galyons]

Good news!!! Enjoy.  Let us know how you like it as it burns in.

Cheers,
Geary