Paradise Builders

Here’s what I built (it was a while ago).

Early on (2013) i had some issues as well, documented in this thread but these were (I thought) resolved. Note for example the additional servo resistors, described here. Also I had to play around with some of the transistor values. Not sure any of this is related to what I'm seeing now though...
 

Attachments

  • IMG_8536.jpeg
    IMG_8536.jpeg
    771.7 KB · Views: 71
I guess you connected the outputs to the scope. Is the input terminated or connected to the signal generator ?
The supply is not shown and I see no connection of signal ground to chassis.

Check integrity of all supply voltages with DMM and try to detect in which branch the oscillation occurs or has
max. level. A scope probe can be used for this.

I also guess that you had to "play around with some of the" resistors values, not "transistor values".
We need the complete picture for additional advice.
 
Need to dig deep to 2013...
The final word on this:

-Q104/204 is a bit hit and miss. I ended up with both around Hfe=460, although it is not a sure thing it will work with that value (I tried multiple with similar Hfe and some gave oscillation and some not)

-I had to remove the caps across Q104 and Q204 (see assembly guide), otherwise it wouldn't work.

-I had to KEEP the caps across Q102 and Q202 (2.7n, I used 2.2n), otherwise it didn't work.

Output connected to scope, input connected to signal generator (which was not generating a signal when I took the picture of the scope). Input and output terminal GND are connected to the respective GND terminals on the pcb. The pcb is not connected to chassis GND (I did a quick test and connected with alligator clips but it didn't improve the measurements).

Will try to measure a bit later.
 
Did some measurements, having a hard time reconciliating the schematic with the boards (my board has the values of the components but not their numbering).

Voltages:
  • Vminus = -18.2V
  • Vplus = 18.2V
  • Vout = 0V
  • Vin = 0V
  • Vext_riaa = 0V
  • Vcascl = -13.3V
  • Vcasch = 13.4V
  • Test point? = -15.8V
  • Testpoint? = -15.7V
Probing around, I see oscillation when probing:
  • Ext_Riaa
  • SK170 and J310 near the ext riaa
  • R12 (4.7R) and R48 (1M)
  • bunch of resistors marked silk screen as 147k, 147k, 18k2, 18k2 (values not present in the schematic). Guessing R43a, R43b, R13, R15
 
In the schematic you see dc voltages referenced to ground. So you can compare yourself.

For instance at the RIAA test point yo should see the gate voltage of the output buffer, indicated
as -1,5 V. Re your probe test points I have to admit I can not read the small designations in the
schematic.

Somebody else may be able to help now. Try to investigate the channels independently and see
what happens if one is disconnected from supply.

The messy diagram has been criticised before - why does it have voltages in big bold letters and parts
designators unreadable ?

The chassis should always be connected to signal ground (not power cord ground), please note that
these anodised sheets are likely to be well isolated from each other. Ground connections require screws
and tooth washers. Alligator clips are not likely to connect properly in this case.
 
Did some more digging, decided to first try to sort out grounding issues before looking for any oscillations that may (or not) remain. Redid some wiring and tried to implement some of the grounding concepts as per this thread. This seems to have improved matters audibly already but some humming remains on one channel.

Measurement shows on the B channel a 90mVpp sinusoid at more or less exactly 50hz. So clearly ground hum. On the A channel also a 50hz sinus but much lower (could be triggered by the issues on the B channel?).

My current wiring, using a 2-box enclosure setup (psu+prereg, amp)
  • separate mains gnd from audio gnd using a 10R resistor.
  • mains gnd not yet fully sorted as I’m testing with the lid open
  • audio gnd for both channels is connected to mains gnd (via said 10r) only at a single point: in the psu enclosure the gnd at the output of the prereg.
  • Other than it that point, audio gnd never meets the chassis or mains gnd

Both channels are connected exactly the same, and for testing the psu box has been placed as far away as possible against EMI.

Not sure what I could be missing here?

IMG_8551.jpeg
 
This is good news. The lf oscillation is gone (if it was in fact 70 Hz as stated in post 4437 / 4439).
But it is not indicated in your scope picture : is the amp input open or shorted?

The mains transformer seems to be separated in a different chassis (I asked before) so it is not
necessary to connect the amp chassis to mains ground.

Do not directly (0 ohms) connect mains ground and signal ground at any point. Consider that the
scope input ground is connected to mains ground (this is true for most scopes, check again with
the Philips you use).

In the end this 50 Hz noise will be removed almost completely ..
 
The picture is with the TT connected. Did some more fact finding today:

  • when I switch around the input from the TT (L/R) the hum moves to the other channel.
  • This indicated a possible issue with my TT, which is a Michel Gyrodec. I spent the whole day trying to figure out any grounding issues within the TT but couldn’t find any.
  • I have a rega RB300 arm, which provides gnd through its left RCA shield - guess what, the hum is on that side. I did test for continuity of gnd, it’s there.
  • The hum goes away when I unplug the motor of the turntable. This is an AC motor with a simple (Not grounded) psu.
  • when I earth the motor through the tonearm, hum turns to loud buzzing.
  • when I connect a separate gnd cable from the preamp to the motor, hum nearly goes away. However this is not the way to do it. It’s strange as well as the motor is not electrically connected to the rest of the turntable. It’s a free standing motor….This really puzzles me….

So now I’m concluding that there is some ground loop created by the TT motor and the preamp…. But I don’t know why as the motor is not electrically connected to the rest of the turntable and I don’t know yet how to get rid of it….
 
Considering the turntable this may be a wiring difference, disparity in twisted pairs and the like.
Both pick up connections should be isolated from turntable ground and all metal parts of the
turntable connected to this tt ground. Check also the Michell is wired to mains safety ground or not.
The possible wiring problem of the Rega arm should be resolved (by using a regular arm). You are
already on the right way.
 
Member
Joined 2005
Paid Member
Paradise is a transimpedance amplifier, so you should look at y_fs (the forward transconductance) of the transistors, the closer the better. Alternatively, beta (the current gain) e.g. at a collector current of 1mA should be a good proxy. Some people had matched for beta and base-emitter voltage at the same time, which is even a little better.
That said, to get 16 matches (for stereo), of which 8 NPNs and 8 PNPs, depending on the precision you are looking for, can be difficult :D
 
  • Like
Reactions: 1 user