Noise from 48V in THAT1512 preamp

[SirEgno]

Hi all,
I made a very nice preamp using THAT 1512/1646 chips. The schematic is the standard one with minor changes. I gave it to a friend and he's using in his studio recording great musicians.
BUT... another friend is experiencing some noises with 48V. The noise appear only with some microphones (e.g. neumann u87) or without any mic plugged.
The power supply use an external transformer (2x18/2x650mA), 7815/7915 for the chips and the schematic taken from the G9 for the 48V. I modified the schematic by making it doubler and not tripler and replaced the trimmer with two 24v zener in series.
Simulating in LT Spice the 48V should be perfect, but in real life it produces some noises. The noise is like pops and cracks, not hiss or hum (I attached a sample whitout mic plugged, I don't have a u87, and 70dB gain).
I tried also the original G9 schematic, but the noise is the same.
I don't know what to do or what to check.
Unfortunately I don't have an oscilloscope.

Any idea?

Thanks in advance!

PS. sorry if my english is not ok, I'm italian and so I'm better at eating then learing foreign languages.

 

[gyraf]

..could be problems in the input-electrolytics of your preamp circuit?

Because of input cmrr, phantom power noise has to be rather bad to break through to audio

/Jakob E.

 

[SirEgno]

Hi Jakob, thanks for answering.
The capacitors are 100uF with positive lead toward the microphone input. I didn't see any schematic with this large capacitors, maybe the problem is over there? I will try to change with a smaller value later.
Due to the fact that the noise appear only with the phantom power engaged I blamed the 48V supply.

 

[gyraf]

if plus's are towards mic amp input, then you're applying +48V at the minus of the electrolytics? please don't do this..

 

[SirEgno]

No the opposite. An image will speak better than me

 

[ccaudle]

the schematic taken from the G9 for the 48V.

I am not familiar with "G9," but that schematic has all of the noise of the Zener diodes applied directly to the base of the capacitance multiplier transistor Q1.
Usually a capacitance multiplier will have the base fed by an RC circuit (that is, a resistor between D5 and C4 to form a low pass filter for the noise).

 

[SirEgno]

I swapped the input capacitors with 47uF. Now it is as silent as the hill.


Thank you Jakob.

 

[Zeltec Service]

I Agree with ccaudle above. I would have used an LM317 with the feedback to the base from the output side. that would eliminate R1,2,4, D4,5,6, and C4. I would put a 200 to 250 Ω resistor from the out pin 3 to the Base pin and another resistor from the base to ground the value of which will determine the output voltage. No cap is necessary on pin 2 (Base). This also provides a bleeder path when turned off. A cap can be applied to pin 2 for a turn on/off ramp. The output cap, C5 would best be a WIMA 10µf @ 63 volts multi layer film cap, very low impedance. I might leave the last cap in the Tripler a higher value. A .1µf multi layer ceramic cap on the in and out pins to ground will suppress any oscillations. This approach would also compensate for the load whether you were powering one or ten mic's with this supply.

 

[SirEgno]

The LM317 is rated for max 40V. I found the TL783, that could be used in the same way, but with higher voltage.
Am I right?

EDIT: the datasheet says that there is a minimum output current to maintain the regulation. 15mA. I think this is less than the maximum absorbtion when all 4 channels are working.

 

[Zeltec Service]

That means that 40 volts between the input and output voltage the pin 2 voltage voltage will never be more than a few volts different than the output voltage. So say there's 55 volts on the input pin 48 volts on the output pin and 46 volts on pin 2 (adjust pin) there's never more than a 12 volt difference across the any part of the LM317. the current is maintained by the 2 resistors that set the voltage.

 

[SirEgno]

The datasheet says "Input voltage range" rated for 4,25 - 40, and in to out differential voltage 3-40V.
I think that there are two different limits at 40V.
https://www.ti.com/lit/ds/symlink/lm317.pdf

I've never used the 317, always 78xx and 79xx, much easier, with less components. I need to study the datasheet.

 

[offkey_99]

My memory, which might be wrong, is telling me that the small-signal voltage noise at the output of an LM317 regulator is lower than that of the 78xx. In many circuits (having large power-supply rejection) this won't matter much but there are circuits that pass on supply noise via biasing networks, etc. and the lower noise would be a benefit.

Also, there is the LM317HV which can handle a higher input voltage.

 

[My3gger]

You don't need additional load for the TL783 with 4k7/2W resistor as shown in schematic by Ian.

 

[Zeltec Service]

You don't need additional load for the TL783 with 4k7/2W resistor as shown in schematic by Ian.

Yes that works also. One thing about the cap on the feed back. If it's too large, it might exceed the device maximum while ramping up. That exact same circuit with those values should work with an LM317 as well. The 4K7Ω need only be ½ watt. As I said that resistor chain provides the minimum current for regulation.

 

[SirEgno]

Thanks all. I'll try this schematic although I don't need mutch than two microphone in a unit.

 

[Zeltec Service]

As far as noise is concerned, the mic pre phantom resistor matching is vital. I have a bunch of 6K8 1% that are matched in pairs to the decimal. The Pre amp's common mode rejection will pass none of that if perfectly matched resistors are chosen. Note to all Neve 80 series console owners. For some unknown reason most of these consoles were sent out with 15KΩ resistors. Under powered mic's react badly to this with noise and distortion in many cases.

 

[Zeltec Service]

If the

As far as noise is concerned, the mic pre phantom resistor matching is vital. I have a bunch of 6K8 1% that are matched in pairs to the decimal. The Pre amp's common mode rejection will pass none of that if perfectly matched resistors are chosen. Note to all Neve 80 series console owners. For some unknown reason most of these consoles were sent out with 15KΩ resistors. Under powered mic's react badly to this with noise and distortion in many cases.

IF the 48 volt card is installed in the meter bridge of your 80 series, all mic are powered. So if you wish to add individual phantom switches and 6K8 resistors the 15K's must be removed. You shouldn't have both in the circuit.

 

[Zeltec Service]

The datasheet says "Input voltage range" rated for 4,25 - 40, and in to out differential voltage 3-40V.
I think that there are two different limits at 40V.
https://www.ti.com/lit/ds/symlink/lm317.pdf

I've never used the 317, always 78xx and 79xx, much easier, with less components. I need to study the datasheet.

I typed a reply but it got lost somehow. The 78/79 devices are pin 2 zero volt devices so the device maximums are also the circuit maximums. This is not true for the LM317/337 and like devices.