trans4funks1
Well-known member
- Joined
- Feb 4, 2013
- Messages
- 328
Nice work!
FEELER - Universal Tube Microphone PSU
- Thread starter Matador
- Start date
Help Support GroupDIY Audio Forum:
chunger
Well-known member
tried connecting proto PSU to mic tonight (C12 clone). . . and results were not good. I had an initial voltage spike upon power up on the B+ to upwards of 200V. I shut it down and put 180K test resistor between pin 1 and pin 7. Lowest adjust value is 134V in current configuration. I'm assuming the current draw of the C12 mic is just a miniscule fraction of the previous test load.
The next revision PCB has a pot in the R3 position so I'm thinking both stages will be able to be balanced and we should have more trim range.
For now, I'm just trying to get the proto up on the C12 so I can test for sonics.
The next revision PCB has a pot in the R3 position so I'm thinking both stages will be able to be balanced and we should have more trim range.
For now, I'm just trying to get the proto up on the C12 so I can test for sonics.
chunger
Well-known member
Knocked R3 down to 33.1K and reduced the B+ under 180K test load to 125V at minimum setting. I deemed this close enough to plug a mic in to take a listen.
We have a persistent non-waivering hum in the sound now sortof a mid pitched "BZZZZZZZZZZZZZZ" sound. Underneath that hum, I hear a functioning microphone with patterning working correctly.
We have a persistent non-waivering hum in the sound now sortof a mid pitched "BZZZZZZZZZZZZZZ" sound. Underneath that hum, I hear a functioning microphone with patterning working correctly.
Matador
Well-known member
Thanks Chunger!
I suspect the buzz sound is 120Hz ripple: can you confirm this is the case on a scope?
I also suspect the problem is with minimum load current. Unlike the LM317T, the TL783C requires a much higher minimum load current to guarantee load regulation specs on the datasheet. If you draw less current, then the ripple rejection can go to shit. Tube mikes generally draw less than 1mA, so they don't guarantee the minimum load current by any means.
LM317T specifies 3.5mA typical, 12mA maximum. TL783C specifies only a maximum of 15mA. So a 221R voltage sense resistor on a LM317T guarantees a minimum of 5.6mA, so that is good for those designs. However I used the same value on this design with a TL783C, which isn't enough.
I changed the latest BOM to utilize 82.5R voltage set resistors, which provide a minimum load current of 15.2mA. I also added a final RC filter stage to the last regulated output, which degrades regulation slightly however adds another filter pole at 100R/47uF = 33.86Hz, which means that the filter is down an additional 12dB @ 120Hz.
I suspect the buzz sound is 120Hz ripple: can you confirm this is the case on a scope?
I also suspect the problem is with minimum load current. Unlike the LM317T, the TL783C requires a much higher minimum load current to guarantee load regulation specs on the datasheet. If you draw less current, then the ripple rejection can go to shit. Tube mikes generally draw less than 1mA, so they don't guarantee the minimum load current by any means.
LM317T specifies 3.5mA typical, 12mA maximum. TL783C specifies only a maximum of 15mA. So a 221R voltage sense resistor on a LM317T guarantees a minimum of 5.6mA, so that is good for those designs. However I used the same value on this design with a TL783C, which isn't enough.
I changed the latest BOM to utilize 82.5R voltage set resistors, which provide a minimum load current of 15.2mA. I also added a final RC filter stage to the last regulated output, which degrades regulation slightly however adds another filter pole at 100R/47uF = 33.86Hz, which means that the filter is down an additional 12dB @ 120Hz.
Phrazemaster
Well-known member
Hi, any updates on this project? I couldn't find the PCB in the store or white market. Or is this the PSU PCB that you are using for the C12 PCB kit?
Thanks; I am considering making an M49 but since my power supply fluctuates a lot where I live, I wanted a regulated design. This is regulated and would work for an M49, right? From what I could tell...
Thanks,
Mike
Thanks; I am considering making an M49 but since my power supply fluctuates a lot where I live, I wanted a regulated design. This is regulated and would work for an M49, right? From what I could tell...
Thanks,
Mike
chunger
Well-known member
Phrazemaster said:
Your timing is perfect. . . Matador is right now have a new (hopefully final) revision board printed for testing. We should have more information soon.
Matador
Well-known member
Indeed! I have a new version of the board with a new regulator circuit for the HV section in the fab for some samples as I type this.
I verified operation of the new regulator circuit in a small prototype panel some time ago, so my hopes are high!
(I just jinxed it, didn't I)
I verified operation of the new regulator circuit in a small prototype panel some time ago, so my hopes are high!
(I just jinxed it, didn't I)
Phrazemaster
Well-known member
Hey Great News! Thanks Matador and Chunger!
Sorry if this is a stupid question, but I've heard some people eschew using regulated PSU's for their mics because they feel that the regulation adds noise. I made a previous M49 with one of Dany's built-like-a-tank RC-RC-RC passive PSU's, but I'm wondering if a regulated PSU might be noisy in some way? Or could it be a drop-in replacement? I would need B+ around 116V and H+ around 4v (using an AC701k!).
Would I need the choke used in Dany's design for better smoothing/noise suppression? I'm not up on how the choke is necessary to begin with...
Problem with the last PSU is since it's passive, and the electricity wanders where I live, sometimes the H+ was going from 3.88v to over 4.2v (bad news for a real AC701K tube!).
Thanks for any feedback/thoughts; it looks amazing and I'd love to be able to use it. Sorry for stupid questions.
Thanks,
Mike
Sorry if this is a stupid question, but I've heard some people eschew using regulated PSU's for their mics because they feel that the regulation adds noise. I made a previous M49 with one of Dany's built-like-a-tank RC-RC-RC passive PSU's, but I'm wondering if a regulated PSU might be noisy in some way? Or could it be a drop-in replacement? I would need B+ around 116V and H+ around 4v (using an AC701k!).
Would I need the choke used in Dany's design for better smoothing/noise suppression? I'm not up on how the choke is necessary to begin with...
Problem with the last PSU is since it's passive, and the electricity wanders where I live, sometimes the H+ was going from 3.88v to over 4.2v (bad news for a real AC701K tube!).
Thanks for any feedback/thoughts; it looks amazing and I'd love to be able to use it. Sorry for stupid questions.
Thanks,
Mike
Matador
Well-known member
The high-voltage design mixes passive filtering with active regulation: there is a CRC filter pre-regulator, and a small RC filter post regulator. The Zener reference is also heavily bypassed, so it shouldn't be any worse than an LM317 circuit (in some ways it should be better). Line regulation sim's out at better than 1%, so a 10% swing on the wall voltage shouldn't move a 116V output by more than 1 volt or so. Since there's an RC filter post-regulator, it will be sensitive to current changes, but when used with a fixed microphone it won't matter (and even if you tried to swap a supply between different microphones, all that would be needed is some turning of an adjustment potentiometer in the worst case).
The heater regulator is based on the previous C12 implementation so it's using a vanilla LM317. LM317's are pretty accurate once they are trimmed out (accuracy is pretty terrible as the LM317's internal VREF has a 10% difference from part to part).
We'll have to characterize the prototype to see definitively. 8) The output is adjustable between about 80V and 240V (up past 50mA or so), so your operating point would be fine.
The heater regulator is based on the previous C12 implementation so it's using a vanilla LM317. LM317's are pretty accurate once they are trimmed out (accuracy is pretty terrible as the LM317's internal VREF has a 10% difference from part to part).
We'll have to characterize the prototype to see definitively. 8) The output is adjustable between about 80V and 240V (up past 50mA or so), so your operating point would be fine.
Phrazemaster
Well-known member
Thanks much Matador; I totally appreciate your sharing so in-depth, and so quickly!
The M49 "b" circuit is susceptible to line noise in the heater circuit H+. With this design, will there be much noise in the H+ line? You didn't mention whether there is filtering post LM317T...I'd really rather go with your design, since my previous PSU is so unpredictable with line voltage swings at the wall.
Thanks!
Mike
The M49 "b" circuit is susceptible to line noise in the heater circuit H+. With this design, will there be much noise in the H+ line? You didn't mention whether there is filtering post LM317T...I'd really rather go with your design, since my previous PSU is so unpredictable with line voltage swings at the wall.
Thanks!
Mike
Matador
Well-known member
A very short update: latest version of the complete PSU board came in last week, and I successfully stuffed and tested the new B+ regulation circuit. It seems stable and working properly.
I re-used a lot of spare parts from other designs, so I need to build two final versions from the real 'official' BOM's and I'll post the test results here. I'm also sending a few copies to Chunger to test as well.
I re-used a lot of spare parts from other designs, so I need to build two final versions from the real 'official' BOM's and I'll post the test results here. I'm also sending a few copies to Chunger to test as well.
Phrazemaster
Well-known member
This is great news Matador!
Thx
Mike
Thx
Mike
Was there any progress on this project? Did I miss the boat or is it on its way?
Bump.
Matador
Well-known member
Well after multiple years and two more versions, I finally arrived at a schematic and BOM that solves all of the intended problems.
The curse of the last three versions was current limiting: it is difficult to implement user-adjustable current limiting around a regular three terminal regulator without degrading ripple rejection. Obviously most 3-terminal regulator have built in current limiting, but I wanted to allow arbitrary current limits that make sense for simple tube circuits like the C12.
Here is the final prototype - at it's core is an LR8 high voltage trimmable voltage regulator, however it is just used as a voltage reference to feed a high voltage MOSFET in source-follower mode. This make it easy to supply more than 20mA of current at widely varying output voltages because the brunt of dealing with high voltages falls to the MOSFET, which can sustain 800V from source to drain.
Here is a pic of the high-voltage section:
The output can be trimmed between 48V and ~280V. Current is limited only by the maximum dissipation in the MOSFET at the set output voltage. I tested the three common use cases for microphones:
1) U67 - 215V output at 4mA (50K test resistor, 1W)
2) C12 - 120V output at 2mA (60K test resistor, 1/4W)
3) U47 - 105V at 50mA (2K2 test resistor, 5W)
U47 is the worst case, as it has the highest current requirement at the lowest voltage, so dissipation is roughly 8W. However the stock heat sink has a theta-j of about 6 degrees C per W, which means around 50 degrees C above ambient, which although hot to us is well under the thermal limit of the MOSFET.
Here is a scope plot of the design trimmed to C12 conditions, loaded with a 60K test resistor:
Output is clean, and ripple is below the voltage threshold on my standard scope probe.
Default current limit set on the BOM is 70mA, so I threw on a 300 ohm, 10W power resistor. Limited to 70mA, maximum voltage output under current limit should be roughly 20V, which jives with what the scope shows:
Despite the MOSFET being rated at 50W, this puts 18W into the MOSFET, so it gets HOT. However the current limit is meant to be a safety issue, and not a permanent mode of operation. I also confirmed the supply can survive into a dead short, at least until the MOSFET heats up and cooks itself (should take at least several minutes, which is enough to realize something is wrong and disconnect it).
I ordered final production boards and I'll pass to Chunger for one of his build threads.
The curse of the last three versions was current limiting: it is difficult to implement user-adjustable current limiting around a regular three terminal regulator without degrading ripple rejection. Obviously most 3-terminal regulator have built in current limiting, but I wanted to allow arbitrary current limits that make sense for simple tube circuits like the C12.
Here is the final prototype - at it's core is an LR8 high voltage trimmable voltage regulator, however it is just used as a voltage reference to feed a high voltage MOSFET in source-follower mode. This make it easy to supply more than 20mA of current at widely varying output voltages because the brunt of dealing with high voltages falls to the MOSFET, which can sustain 800V from source to drain.
Here is a pic of the high-voltage section:
The output can be trimmed between 48V and ~280V. Current is limited only by the maximum dissipation in the MOSFET at the set output voltage. I tested the three common use cases for microphones:
1) U67 - 215V output at 4mA (50K test resistor, 1W)
2) C12 - 120V output at 2mA (60K test resistor, 1/4W)
3) U47 - 105V at 50mA (2K2 test resistor, 5W)
U47 is the worst case, as it has the highest current requirement at the lowest voltage, so dissipation is roughly 8W. However the stock heat sink has a theta-j of about 6 degrees C per W, which means around 50 degrees C above ambient, which although hot to us is well under the thermal limit of the MOSFET.
Here is a scope plot of the design trimmed to C12 conditions, loaded with a 60K test resistor:
Output is clean, and ripple is below the voltage threshold on my standard scope probe.
Default current limit set on the BOM is 70mA, so I threw on a 300 ohm, 10W power resistor. Limited to 70mA, maximum voltage output under current limit should be roughly 20V, which jives with what the scope shows:
Despite the MOSFET being rated at 50W, this puts 18W into the MOSFET, so it gets HOT. However the current limit is meant to be a safety issue, and not a permanent mode of operation. I also confirmed the supply can survive into a dead short, at least until the MOSFET heats up and cooks itself (should take at least several minutes, which is enough to realize something is wrong and disconnect it).
I ordered final production boards and I'll pass to Chunger for one of his build threads.
Matador
Well-known member
Installed in the PSU case.
Brilliant! I need about 6 of these. Patiently waiting...Installed in the PSU case.
