Author Topic: QSXM2 - various debugging  (Read 7296 times)

0 Members and 1 Guest are viewing this topic.

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
QSXM2 - various debugging
« on: April 14, 2014, 10:17:52 PM »
Hi,
I was following the cautious way of building the QSXM2.
The QSXM2 uses one pair of SSM2210/2220 per channel in positons T5+T23/T6+T24. I use 3,6 kOhm resistors in R16 and R17. In all other locations I use BC550C and BC560C, manually matched hfe so that the pre-amp has hfes between 565-574 for the 560C and 571 to 578 for the 550C. Supply voltage is +/-15V. I did not place a jumper at J1, so that I could somewhat compare the measurements to those under "Pre amp, Left channel, MM-mode, BC550C, BC560C in all positions, OP07 as DC-servo, no input bias servo, input not shorted."

Measurements look fine, but there are some that are "off":

Reference measurement|Measurement, left channel|Measurement, right channel
R27, 9,15V, 91,5uA6,33V6,31V
R28, 3,65V, 36,5uA6,33V6,32V
R29, 9,12V, 91,2uA6,34V6,34V
R30, 3,62V, 36,2uA6,34V6,32V
R38, 834mV, 14,9uA2.08V1.95V
R39, 855mV, 15,3uA1.82V1.92V

Also, the voltage drop across R42 to R47 is in the ballpark of 621 mV to 671 mV for the left channel, while it is 478 to 535 mV for the right channel... For the left channel, this is more than 100% more than the reference measurement.
Another difference is that the voltage drop across diode H1 (left) is 2.04V for the left channel and 1.87V for the right (H3)

Is something alarmingly odd here, or are the aforementioned differences to be expected - caused by the difference in H1 and H3 due to variances in LEDs? Particularly R38 and R39 are also way off - is it due to using the matched SSM2210/SSM2220 transistor pairs?

 

Offline peranders

  • Administrator
  • Reached Nirvana
  • *****
  • Posts: 1230
  • Karma: +6/-0
  • Gender: Male
  • Forum owner
    • Sjöström Audio
Re: QSXM2 - various debugging
« Reply #1 on: April 15, 2014, 06:26:05 AM »
If the collector currents have the expected values and the DC-servo opamp works within its limits then everything is OK.
/Per-Anders Sjöström, owner of this forum

Homepage with my DIY hifi stuff

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #2 on: April 26, 2014, 10:37:28 PM »
And it seems to be the case, so I have progressed onto the RIAA part.

Measurements seems to be fine and very near the same for both channels, but one thing seems to be wrong:
On the left channel, the DC output level quickly goes from several volts, but then it only very slowly progresses to 0V offset. It takes about 5-6 minutes before it reaches 0V.
The right channel starts out at similarly high DC output levels (4.5V or so), but has reached 0V within 30 seconds max.

When testing this, jumpers are placed at J1 and J3 and nowhere else. The DC output on the pre-amp outputs are both 0V.

Do you have any suggestions as to what could be the culprit here? I have checked all solderings multple times, but the result is always the same.
My main thought is that DC shouldn't be present at all on the output during start-up either, but maybe I'm wrong.

Measurements, when both channels have eventually settled:

Code: [Select]
RIAA PART LEFT CH RIGHT CH
DZ5, 6,99V 6,61 6,62
DZ6, 6,98V 6,64 6,61
DZ7, 7,03V 6,8 6,72
DZ8, 6,60V 6,79 6,7
R56, 1,14V, 11,4uA 0,797 0,861
R57, 1,10V, 11,0uA 0,801 0,8
R58, 5,24V, 1,344mA 5,16 5,18
R59, 5,25V, 1,346mA 5,19 5,19
R60, 9,71V, 1,428mA 8,97 8,99
R61, 9,10,V 1,338mA 8,96 9
R62, 5,25V, 772uA 5,22 5,23
R63, 5,28V, 776uA 5,25 5,24
R64, 4,28V, 764uA 4,27 4,24
R65, 4,33V, 773uA 4,27 4,26
R66, 183mV, 1,83uA 0,214 0,22
R67, 209mV, 2,09uA 0,31 0,322
R68, 446mV, 656uA 0,41 0,417
R69, 451mV, 663uA 0,402 0,402
R70, 1,45V, 659uA 1,289 1,292
R71, 1,47V, 668uA 1,262 1,27
R72, 1,31V, 1,191mA     1,134 1,135
R73, 1,32V, 1,200mA     1,11 1,114
R74, 667mV, 6,67mA 0,497 0,498
R75, 665mV, 6,65mA 0,505 0,494
R78, 629mV, 11,23uA 0,494 0,497
R79, 687mV, 12,22uA 0,457 0,447
R82, 338mV, 15,36mA 0,309 0,305
R83, 337mV, 15,32mA 0,306 0,308
C30 5,8 5,8
H2, 1,89V 1,85 1,83

Additional information:

With no op-amps IC2/IC4 in place and 15V at pin 6, the via at the voltage division between R87 and R88 becomes is about -40mV at -15V and +40mV at +15V, which is what it should be.
But when I measure at the output of the amplifier without IC2/IC4 in place, I get the following readings:

Left channel:
+15V at pin 6, it settles at 0.246mV
-15V at pin 6, it never really settles but wants to stay at about -11.20V (yes, volts).

Right channel:
+15V at pin 6, ends at 1.02V
-15V at pin 6, again never really settles but wants to stay at about -11V.

I have looked at the transistor markings (550C and 560C) countless of times and found no errors or mistakes. But I'm thinking that it must somehow be a problem with the transistors (which also are matched by hand).





« Last Edit: April 27, 2014, 02:17:49 PM by tk-don »

Offline peranders

  • Administrator
  • Reached Nirvana
  • *****
  • Posts: 1230
  • Karma: +6/-0
  • Gender: Male
  • Forum owner
    • Sjöström Audio
Re: QSXM2 - various debugging
« Reply #3 on: April 30, 2014, 07:13:38 AM »
Have you checked R85, C54, R87 and R88. Compare those with the working channel. How much DC voltage do you have at R93?
/Per-Anders Sjöström, owner of this forum

Homepage with my DIY hifi stuff

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #4 on: April 30, 2014, 09:32:27 PM »
I have a very hard time registering the transients when powering up the amp and using my own cheap multimeter, so I'll use an oscilloscope at work tomorrow.

What I do notice though is that the voltage at the output of the opamp integrator is generally very high for both channels. When settled: 12.65V for the right, 13.90V for the left channel. When powering up, the left channel integrator op-amp output hits the +V voltage rail  at 14.16V (probably the closest OP07 gets to 15V as it isn't rail-to-rail).

When settled these are the voltages:

Voltage across R85:
+/- a few milivolts for both channels

Approx. voltage across C54:
Left: 13.10V, right: 11.50V

Approx. voltage across R87:
Left: 14.05V, Right: 12.40V

Approx. voltage across R88:
Left: 37.9 mV, Right 35.1 mV

Voltage across R93:
Within 1 mV (+/-) for both .. not sure if all this is just measurement uncertaincy.

Would it make more sense to debug without IC2/IC4 installed?

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #5 on: May 02, 2014, 07:35:32 PM »
Hmm, after testing with an oscilloscope it turns out that:

1) At both TP1 and TP2 I actually have an AC signal/sinusoid at around 275 MHz and 800 mV pk-pk! As R2/E2 I use a ferrite bead on an uninsulated wire. The small ferrite beads were something leftover from a previous project - not sure of the specs, but I suppose it isn't very important.

2) When I touch/push the cluster of resistors R42-R47 with a finger, and measure the output at R85 with respect to ground, the output quickly turns to 0V. When I remove my finger again, the offset at R85 re-appears, but slowly goes to 0V as I described above. When I do the same for E42-E47, nothing happens at R85.... output is consistent.

I'll have a look at my solderings in the pre-amplifier left channel again, since something looks wrong. However, should I just ignore the presence of the 275 MHz signal at TP1/TP2? I'm testing the PCB without any enclosure, and with nothing connected to the chassis/gnd inputs, should this affect anything? No matter what components I touch the 275 MHz signal is present, but of course not at the output of the RIAA part.

Thanks!

EDIT: All transistors are mounted with as short leads as possible, max ~2.5 mm is visible.
« Last Edit: May 02, 2014, 08:41:04 PM by tk-don »

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #6 on: May 28, 2014, 09:36:40 PM »
Still debugging on this, haven't had the time to do much experimentation. One question though:

I just found out that the hFE of the SSM2220 and SSM2210 are not even close to each other, and the data sheets also agree with this. It's about 165 for the 2220 and 460-470 for the 2210. But the matching of them per package is fine.
Do any components need to be change to accomodate the fact that they differ so much in gain?

Otherwise I've toyed a bit with increasing the capacitance of C17 and C18 with 22 pF, but without any changes at all, still the 275 MHz remains.
But what I see is that it draws power from the power supply in such a fashion that 275 MHz is generated at the output of the power supply to a very small extend, but this ripple seem to become larger the further away from the power supply input connectors on the QSXM2 pcb that I measure, i.e. it's amplified.
By further away I'm talking about performing measurements between one of the supply line running along the edge or middle or the PCB and then ground.
All of the electrolytic capacitors I use are Elna RJH (low ESR) measured to be at their rated capacitance.

Offline peranders

  • Administrator
  • Reached Nirvana
  • *****
  • Posts: 1230
  • Karma: +6/-0
  • Gender: Male
  • Forum owner
    • Sjöström Audio
Re: QSXM2 - various debugging
« Reply #7 on: May 29, 2014, 07:04:26 AM »
You haven't been able to pinpoint where the source of this 275 MHz is? Is it the SSM's or is this oscillation present everywhere?

The unbalance of the SSM's will create a negative input bias current but this will the bias servo handle.
/Per-Anders Sjöström, owner of this forum

Homepage with my DIY hifi stuff

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #8 on: May 29, 2014, 02:16:49 PM »
I haven't had the time to do many hours of experimentation for the last couple of weeks, but yes, it is basically present everywhere and tested with multiple oscilloscopes and probes. I have tested both with a Hameg power supply (7042 type) and the SSR03. When they are loaded by the QSXM2, 275 MHz is present on the power supply outputs but only by an extremely small amount.
The BC550C at T31,T33 and T35 get somewhat hotter than the BC560C in T32, T34 and T36. The BC550C become too hot to touch for more than 5 seconds after the amp has been on for some minutes.

My guess is to try modifying the location of poles created at various places in the preamp, like  C7/R31, C17/R37 but I'm not sure if this is the right approach to take. Maybe I'll need to modify the value of R50 somewhat too.

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #9 on: June 07, 2015, 08:17:15 PM »
Hi again,

I've had to suspend working on this for a very long time, I just haven't had the time to tear anything apart to find out the source of this ~275 MHz interference. It is on both channels by the same amount.
What REALLY bothers is that when disconnecting R50, thereby also disabling feedback, the amplifier by itself still oscillates at 275 MHz, but to a little lesser degree. So something is seriously not working.

When R50 is in place (68 Ohms):


When R50 is removed:



My question is mainly - have you seen it before, or do you have any good ideas what to do?

A few ideas, that came to mind are:
- Could this be due to bad transistors or simply "mismatch" between manufacturers ? The BC560C are from ON, the 550C are from "Taiwan Semiconductor", but I have matched them in sets, regarding hFE. I couldn't find BC550C from ON Semiconductor at that time, which is why I had to choose something else.
- I'm using sockets for all SSMs - high quality tulip-style sockets
- All capacitors are low esr Elna RJH series. Could the ESR affect anything?

The reason why I didn't experiment with the above issues is that it requires a more drastic amount of work, and I'll need to source a hot plate or soldering heat gun to remove sockets without damaging the pcb pads.


Offline peranders

  • Administrator
  • Reached Nirvana
  • *****
  • Posts: 1230
  • Karma: +6/-0
  • Gender: Male
  • Forum owner
    • Sjöström Audio
Re: QSXM2 - various debugging
« Reply #10 on: June 07, 2015, 08:42:23 PM »
Can you tell me if you have used exactly my BOM except for the brand of the BC's?

Are you sure that the signal isn't coming from outside?

The frequency is a little bit high for coming from feedback.

How much voltage do you have over R42-R47?

The same problem in the other channel?

If have shorted the input?
/Per-Anders Sjöström, owner of this forum

Homepage with my DIY hifi stuff

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #11 on: June 07, 2015, 10:28:40 PM »
Thanks for replying so quicky. Same issue on both channels, input channels shorted to ground.

1)
Differences to the BOM is as follows:

- All 100nF/63VDC film capacitors (65-226-84) are of type polycarbonate (Evox CMK series, 2.5% tolerance, same voltage rating)
- All 10 uF and 100 uF capacitors are Elna RJH series
- OP07 instead of obsolete MAX420, and thus no DZ9, DZ10, C61 and C62

This modification was made in order to supposedly better match the cartridge impedance:
- R1: 2k2 Ohm , bom value 56k
- R2: Ferrite bead - tried with a low value resistor: no change
- R3: 220 Ohm, bom value 100 Ohm

I'm using 3.6kOhm in R16 and R17.

2) I have tested this in free-air and also while it was fully enclosed in a shielded antistatic bag. Measurements are the same, but I don't know if this type of shielding would be sufficient.

I'll get back to R42 and R47, because since I made the experiment with disconnecting R50, after inserting it again, it heats up rapidly and eventually burns up. This experiment with removing R50 - do you think it could have caused this - obviously something is very wrong now. I'd hope to have a look at it again tomorrow.

Offline peranders

  • Administrator
  • Reached Nirvana
  • *****
  • Posts: 1230
  • Karma: +6/-0
  • Gender: Male
  • Forum owner
    • Sjöström Audio
Re: QSXM2 - various debugging
« Reply #12 on: June 08, 2015, 06:26:07 AM »
2k2 for a MM cartridge seems very low.....

Do you have the same problem MM/MC mode?

R43 etc is really 22 ohms?

H1, voltage across it?
/Per-Anders Sjöström, owner of this forum

Homepage with my DIY hifi stuff

Offline tk-don

  • Member
  • **
  • Posts: 28
  • Karma: +0/-0
  • Custom title, please change to something cool.
Re: QSXM2 - various debugging
« Reply #13 on: June 08, 2015, 09:36:21 PM »
Ah, yes - this is for a Hi-output MC cartridge - around 2mV. Recommended is supposedly 2k to 47k.

Seems like i damaged one or more of T31, T33 or T35 when I did the measurements the last time so I'll replace those, so in the meantime the measurements are for the other channel.

Both channels have identical issues, and the problem is the the same regardless of MM or MC mode. These measurements are with IC3/DC feedback in place, using a multimeter - can use a scope as well, but the RMS measurements should be the same.

H3: 1.895V
E42: 0.464V
E43: 0.578V
E44: 0.538V
E45: 0.634V
E46: 0.478V
E47: 0.667V

Measurements differ a bit from time to time - can differ up to about +/20mV
All R42 to R47 and E42 to E47 are 22 ohm... ok, between 22.4 and 22.6 ohm.  :-)
« Last Edit: June 08, 2015, 09:42:23 PM by tk-don »