Hi Salas / group,
I stuffed the DCG3 board with parts from Teabag and have hooked it up to a Salas uBib set at +/-17v. Upon firing up, the negative rail reads -0.5v. With the two sides powered separately, it reads -1.5/-1v. I wonder if anyone could speculate what might cause this. I fear I have populated something incorrectly on the negative side of the DCG3, or that something is wrong with M3 (test as resistors in a multi-tester). The three LEDs fire up and the switch clicks. The positive rail reads 17v. Any suggestions?
Cheers, Soren
I stuffed the DCG3 board with parts from Teabag and have hooked it up to a Salas uBib set at +/-17v. Upon firing up, the negative rail reads -0.5v. With the two sides powered separately, it reads -1.5/-1v. I wonder if anyone could speculate what might cause this. I fear I have populated something incorrectly on the negative side of the DCG3, or that something is wrong with M3 (test as resistors in a multi-tester). The three LEDs fire up and the switch clicks. The positive rail reads 17v. Any suggestions?
Cheers, Soren
Greetings....
I'd like to build a second DCG3 (still enjoying the first build with Ultrabib's) but this time I'd like to build a two chassis solution. Something similar to what Jaytor did in his DCG3 build but SE not BAL. I'd like to have a separate chassis for the power supply and connect via umbilical to the main pre-amp chassis which will have Ultrabib's. I'd like to have some initial power filtering performed in the PSU chassis before it reaches the main pre-amp chassis.
So here are my questions:
1) Would it be a good idea to use the Salas DCSTB as an initial filter in the power supply chassis to feed the Ultrabib in the second chassis? I realize I would need four.
2) Similarly could I use two Ultrabib's in the power supply chassis to feed the Ultrabibs in the second chassis?
Would #2 be a better solution than #1?
I already have DCSTB' s (built) and extra Ultrabib's boards (not built up)
If these are bad idea's can someone recommend a linear power supply board that can perform the initial filtering before the power reaches the main chassis where the Ultrabib's will be?
Thanks in advance.
Frank M
I'd like to build a second DCG3 (still enjoying the first build with Ultrabib's) but this time I'd like to build a two chassis solution. Something similar to what Jaytor did in his DCG3 build but SE not BAL. I'd like to have a separate chassis for the power supply and connect via umbilical to the main pre-amp chassis which will have Ultrabib's. I'd like to have some initial power filtering performed in the PSU chassis before it reaches the main pre-amp chassis.
So here are my questions:
1) Would it be a good idea to use the Salas DCSTB as an initial filter in the power supply chassis to feed the Ultrabib in the second chassis? I realize I would need four.
2) Similarly could I use two Ultrabib's in the power supply chassis to feed the Ultrabibs in the second chassis?
Would #2 be a better solution than #1?
I already have DCSTB' s (built) and extra Ultrabib's boards (not built up)
If these are bad idea's can someone recommend a linear power supply board that can perform the initial filtering before the power reaches the main chassis where the Ultrabib's will be?
Thanks in advance.
Frank M
Hi,
-Study your tester's spec to see if it can recognize a DMOS. If it can but sees M3 as a resistor then it must be a broken M3 creating a short.
-If the Ubibs give +/-17V on their output wires before connected to the DCG3 but then the -V goes far down, maybe current limiting occurs.
This can be because the negative DCG3 rail pulls over-current and the minus Ubib's current source thankfully intervenes. It can either be due to true fault current or you set the negative Ubib too conservatively for the bias current of your DCG3. In this case recalculate R1 in the associated Ubib section. In any case check that no MOSFET shows continuity from its metal tab to the sink plate. Also check that no MOSFET type has been mixed up for placement in the Ubibs and/or the DCG3.
fmena, you can have a simple unregulated supply with big filter capacitors (hence little ripple) in the first chassis.
DCSTB is a capacitance multiplier so if I had to experiment with other power supply, I would go to the standard linear regulators (maybe as a tracking regulator before UBIB). The VRDN implementation of our trusty LM317 is a good candidate. You can also use it instead of UBIB if you like.
I wouldn't use Super Regulator or another shunt reg in front of UBIB and since you already have DCSTB, can't see the point of another Cap Multiplier.
DCSTB is a capacitance multiplier so if I had to experiment with other power supply, I would go to the standard linear regulators (maybe as a tracking regulator before UBIB). The VRDN implementation of our trusty LM317 is a good candidate. You can also use it instead of UBIB if you like.
I wouldn't use Super Regulator or another shunt reg in front of UBIB and since you already have DCSTB, can't see the point of another Cap Multiplier.
Last edited:
Hi,
Not very good ideas. You could instead set DCSTB at +/-22V* as separate box pre-regulators and feed DC to Ubib sections with a umbilical at their C1s pads underneath PCB. Use no rectification diodes on Ubib and 100uF C1s. Set +/-17V Ubib output. If you get something like +/-25V raw DC or more at the DCSTB main filter caps with your available transformers its doable. Only see that the Schottky BOM rectification diodes don't heat up too much on DCSTB with Ubibs following, else use thicker ones. You may also increase the DCSTB standard fuse ratings if they will tend to blow.
*You basically increase the Rxj Ryj Ω values for 5V additional drop on them (17V+5V=22V). After you find how many mA go to each one. Ohm's law indicates mA i.e. by Rx,Ry individual Vdrops divided by their now Ω value. Should be in the 5mA vicinity. Determined by the CCS JFETs Idss tolerances feeding LEDbars thus not bang on.
Salas and Savvas - thanks for the suggestions. I did not test with a resistive load, yet. On a quick glance at my notes, the uBib has R1 = 2.2R so ~270mA bias. The DCG3 has R10 = 7.5R which seems to be something like 140-150mA. If my math is right, I would have thought that to be enough overhead. No continuity with the sink.
Yeah, you were powering DCG3 with only one UBIB. Also remember that DCG3 draws little current to the negative side so R1 should have larger value in the negative side (less bias)
Follow the instructions given by Salas for DCG3 powered by one UBIB or dual UBIBs. For more information and explanation by Salas look here
Follow the instructions given by Salas for DCG3 powered by one UBIB or dual UBIBs. For more information and explanation by Salas look here
@Sorenm
Congrats for running the whole setup normally now. Current limiting function in the Ubibs is a welcome safety side effect. It can also intervene when a specific current set limit was forgotten in an experiment. Just like how a bench PSU would act in CC protect mode. Positive CC is set more potent for a DCG3 because it can peak higher there. How low the negative Ubib's constant current was set by measurement (R1 Vdrop/R1)?
Congrats for running the whole setup normally now. Current limiting function in the Ubibs is a welcome safety side effect. It can also intervene when a specific current set limit was forgotten in an experiment. Just like how a bench PSU would act in CC protect mode. Positive CC is set more potent for a DCG3 because it can peak higher there. How low the negative Ubib's constant current was set by measurement (R1 Vdrop/R1)?
I love it! It's a keeper and I am working on putting the lid back on the preamp. A minor space management challenge🙂
I noticed the bridge pads under the PCB, which seem to be for joining gnd between the two sides. Currently, audio gnd is joined at the RCA in/out and shared between the two channels. However, it would be possible to separate audio gnd between the two sides from input to output. Would this be preferrable? If not separated, should the pads under the PCB be joined?
Cheers, Soren
I noticed the bridge pads under the PCB, which seem to be for joining gnd between the two sides. Currently, audio gnd is joined at the RCA in/out and shared between the two channels. However, it would be possible to separate audio gnd between the two sides from input to output. Would this be preferrable? If not separated, should the pads under the PCB be joined?
Cheers, Soren
Hi, congratulations. Nice that you like it. What is the rest of your system in brief?
If not separate at the RCAs its the same as being joined at the PCB channels ground bridge i.e. take no further action.
Fully separate is another test option when facing hum & buzz problems in a system.
Without high resolution spectrum measurements we can't say which one approach is few dB better for harmonic noise in various complete audio systems ground maze. Unless one is audibly quieter than the other.
If not separate at the RCAs its the same as being joined at the PCB channels ground bridge i.e. take no further action.
Fully separate is another test option when facing hum & buzz problems in a system.
Without high resolution spectrum measurements we can't say which one approach is few dB better for harmonic noise in various complete audio systems ground maze. Unless one is audibly quieter than the other.
Hi Salas,
Thanks for the insights. Given there are no audible noise issues, I will keep it simple/as is.
The rest of the system includes the following:
I will share a photo once I get the innards organized 😊
Cheers, Soren
Thanks for the insights. Given there are no audible noise issues, I will keep it simple/as is.
The rest of the system includes the following:
- Miro PCM63 DAC
- DCG3 (incl. Khozmo shunt) - DCG3 replaced another 3x/buffer stage and is much more to my liking
- XA252 (ZM)
- OSMC speakers (mbrennwa)
I will share a photo once I get the innards organized 😊
Cheers, Soren
Here is the current state-of-affairs. The DCG3 is biased around 150mA and powered by two shunts. Input/volume is handled by a Khozmo shunt/board.
The DCG3 sounds lovely. It has been a little over 100 hours, and it throws a large sound stage and with a natural tone/timbre.
Thank you Salas!
The DCG3 sounds lovely. It has been a little over 100 hours, and it throws a large sound stage and with a natural tone/timbre.
Thank you Salas!
Quick question. Thinking about heat management ... I currently have R10 = 7.5R which is ~155mA bias. Would switching to 10R here, lowering the bias/heat a little, impact the sound? This is used as pre amp only. Just wondering, not sure if I want to 'dial it down' 😆
Hi, as line preamp only, dialing the bias down to 100mA does not impact measured spec. It should not impact the line out sound either as I remember from when I did such tests. Your impressions may slightly vary. High bias is basically for better THD spec with 32-80Ω headphones. To fully benefit heat-wise you should also dial down the ZM shunt regs accordingly.