THAT1512 / Transformer 'blind test' mic preamp

[richiyobs]

try to add 22pf on both R32 and R33 of your schematic... i had same problem, then i found it on any console schematic and it seems to work
sine is 10Khz

 

[jamesk]

After not having a chance to work on this all week, due to work, I've done a bit more investigating. I plotted the frequency response of the differential output signal and it seems like it's perfectly flat enough.



I found an interesting article here:

https://www.eetimes.com/cross-coupled-output-stages-for-balanced-audio-interfaces/

Particularly this point:

"Especially in discrete balanced output stages, some designers trim signal balance by using a pot connected between the two outputs with its wiper grounded. (See Figure 2 .) But, the best CMR of the source/receiver system occurs at a very different setting from that which achieves best signal balance. "

Unfortunately there isn't any mention of how to adjust the trim-pot to achieve the best CMRR, and I'm not sure how you'd do that.

Thanks,
James

 

[jamesk]

I have a bigger problem now, though. I finished wiring up the power supply and it's not working. I'm measuring the following voltages:

Transformer secondary: 30V peak
Rectifier +ve: +49V
Rectifier -ve: -30V
+17V output: +48V
-17V output: -28V
+48V output: -7V

I'm pretty stumped. I can't see any problem with the schematic and I've checked all the components. Any ideas?

Here's the schematic again:



Thanks,
James

 

[OneRoomStudio]

I have a bigger problem now, though. I finished wiring up the power supply and it's not working. I'm measuring the following voltages:

Transformer secondary: 30V peak
Rectifier +ve: +49V
Rectifier -ve: -30V
+17V output: +48V
-17V output: -28V
+48V output: -7V

I'm pretty stumped. I can't see any problem with the schematic and I've checked all the components. Any ideas?

Here's the schematic again:

View attachment 145885

Thanks,
James

Are you testing with a load? It looks like the regulators aren’t regulating. Are you sure R1 & R2 are the correct values? You’d probably be better off with trim pots in those positions. You should also have the equivalent diodes on D5 on the other two regulators to prevent current flowing through the output of the regulator. Also, I recently ran into an issue where a regulator was not the same pinout as I was expecting, which was a head-scratcher.

For your phantom circuit, it seems that you’re taking a tap from after the rectifier and rectifying it further? Also, you’re starting with too low a voltage (you can’t regulate 49V down to 48V successfully).

 

[jamesk]

@OneRoomStudio The phantom regulator is fed from a voltage doubler (functioning as a tripler here, being tied to the unregulated positive rail), formed from C10, D4 and D7. The first thing that's confusing me is why I'm getting the wrong voltage on the +ve side of the regulator. With 30V peak on the secondary, I should be getting approximately 30V DC.

 

[DaveDC]

If this is on a PCB, I would suspect a schematic symbol or PCB footprint pinout error, and the actual circuit is not wired as it appears on the schematic. Double check that the component pinouts agree with the datasheets, particularly the bridge rectifier.

Update: yes, you have pins 1 and 2 swapped on your FWB schematic symbol.



Oh, and your voltage setting resistors appear to be wrong for the output voltages you want. The regulators impress a 1.25V reference voltage between their output terminals and their adjust terminals. This flows through your 100 ohm resistors producing a 12.5 mA current. This current flows through the voltage set resistors R1 and R2 to produce a voltage drop that is 1.25 volts less than your desired output voltage. I think you want 1.26K for R1 and R2 for a 17 volt output.

And at about 0.2 watts, the power dissipation of these resistors is a bit high. You might raise the value of R3 and R4 to maybe 500 ohms or even higher to minimize wasted power.

The +48 regulator has a similar dissipation problem, with R6 consuming about 0.6 watts. You'll want to raise R5 to reduce the power in R6 to something reasonable.

 

[jamesk]

@DaveDC Thanks for the hint about footprints. I checked the rectifier and the pin numbering on the schematic symbol is just a bit odd. The rectifier is wired correctly on the PCB.



I think I've found the problem, though! I'm surprised this happened as I used the built-in DO41 symbol so I didn't consider that it might be wrong, and just soldered everything as it appeared on the silkscreen.



No wonder it doesn't work!

With regards to the resistor values, those were meant to be 1k3, not 4k3, I'm not sure how that happened. I see what you're saying about the power dissipation, though. Is there any reason that I can't just change R3 and R4 to 1k and R1 and R2 to 12k6?

With the +48V regulator, if I set R5 to 1k, I calculate that the total resistance from pin 1 to ground should be 37.4k. If I swap R6 to 33k and RP1 to 10k, that should allow me to adjust it correctly, I think.

Does that all sound sensible?

 

[DaveDC]

That FWB symbol still seems messed up even though it may end up being wired okay. The pin numbering and pin order seem wrong. And there is a double line on the silkscreen on the pin 4 end. Should be a pin 1 indicator. Square pad for pin 1, too.

You should strive to ensure your personal symbol library is correct and tidy. Bad symbols are the #1 cause of board spins in my experience. Never trust the built in libraries.

Have a look at the regulator’s adjust pin current on the data sheet. Organize the voltage setting resistor current to about 10x of that value so that the adjust pin current doesn’t contribute much error to the target voltage. Or better, do a proper KCL node analysis at the adjust pin to include the effect of the adjust pin current on the output voltage. Just make sure you get the direction of that pin current right.

 

[jamesk]

@DaveDC Yes, I agree, it's just lucky that it ends up being correctly wired on the PCB. But yes, I'm going to try and build up a library of correct symbols now.

The maximum adjust pin current for the regulators is 0.1mA, so with 1k on the output pins, I'll get 1.25mA of current, which seems okay. That would give a power dissipation of about 60mW in the voltage setting resistor on the +48 regulator.

 

[Walrus]

Don’t forget that the TL783 regulator requires an output current draw of 15 mA to maintain regulation.

 

[ricardo]

Dem cross-coupled o/p stages, Self's, DRV134 etc emulate a FLOATING output transformer NOT a 'balanced' o/p.

Looking at one side with nothing on the other is meaningless.

But an impedance balanced stage will be a LOT better than any of these ... except when you have to deliver zillion dBm into 600R. Instances when you have to do this on a postcard please.

Difficult to get better THD, noise bla bla than a MO resistor and a good electrolytic cap.

---

BTW, your -20dB pad should be AFTER the transformer and/or AFTER C3/4 ... otherwise your P48V will be wonky. You really shouldn't need a -20dB pad on either circuit with your switched gain.

 

[DaveDC]

@DaveDC Yes, I agree, it's just lucky that it ends up being correctly wired on the PCB. But yes, I'm going to try and build up a library of correct symbols now.

The maximum adjust pin current for the regulators is 0.1mA, so with 1k on the output pins, I'll get 1.25mA of current, which seems okay. That would give a power dissipation of about 60mW in the voltage setting resistor on the +48 regulator.

You dodged a bullet with your FWB footprint, and at least the diode rework is easy! I have found that the libraries in the free ECAD programs (EasyEDA, etc.) are loaded with errors, so always check them carefully. As you gain EE/PCB experience, you will appreciate the value of well designed schematic symbols and PCB footprints. Schematics that are easy to read and understand are essential on large projects involving multiple engineers. And well designed PCB footprints (IPC-7351 compliant) make for good high-volume yield.

 

[DaveDC]

---

BTW, your -20dB pad should be AFTER the transformer and/or AFTER C3/4 ... otherwise your P48V will be wonky. You really shouldn't need a -20dB pad on either circuit with your switched gain.

Have a look here and here and you'll see that the -20dB pad is fine as-is. However if you look here you'll see the first reference uses "Type A" phantom switching. You may wish to use "Type C" switching if your console will get used in live sound applications where mics are "split" by literally connecting them to two consoles simultaneously with a Y cable. Type A will load the phantom if one console has its phantom on and the other off, whereas Type C will not . I learned this the hard way during beta testing a console with Clair Brothers...

 

[richiyobs]

https://www.google.com/url?sa=t&sou...8QFnoECCcQAQ&usg=AOvVaw2oLamhrPpXDJY8ViDk1P7k

This Is Jensen an003 about balanced line ... Fig.2.4 It' s a good example that shows how to do an impedance balanced out with all components needed... On some books it's not complete, not showing the cap to Gnd...
No trim to adjust and good interfacing for both balanced and unbalanced receiver without dB loss

 

[ricardo]

BTW, your -20dB pad should be AFTER the transformer and/or AFTER C3/4 ... otherwise your P48V will be wonky. You really shouldn't need a -20dB pad on either circuit with your switched gain.

My thanks to DaveDC for putting right my brain fart .. for which I can only plead senility. The THAT datasheets are indeed correct.

I urge those using THAT devices to follow their latest datasheets and application notes religiously. There are many minor details which aren't fully explained but make a big difference in performance and particularly reliability.

 

[jamesk]

Don’t forget that the TL783 regulator requires an output current draw of 15 mA to maintain regulation.

Hmm, good point. It looks like I can't up those resistor values then. If I did, I'd be relying on the external load to keep the regulators working. That's a big power dissipation, though, so I'll need to up the resistor power ratings. I wonder if it's still possible to salvage the board. Possibly not!

 

[jamesk]

Actually, maybe I can make it work by mounting the resistors vertically.

 

[DaveDC]

Have a look at the TPS7A4001 which is stable with no load as long as the setting resistors draw at least 10uA. Its 50mA maximum output can supply phantom for three mics. Not sure how many mics you need to power, though.

 

[jamesk]

I've replaced the diodes and the resistors now. The +/-17V outputs are now working fine (outputting just over 17V each), but the problems continue with the +48V output. I'm measuring 85V on the output, 86V on the input to the regulator, and only 0.17V across R5. Any suggestions?

 

[DaveDC]

You reversed D6 as well, right? It too has a backwards symbol.