Most likely reason (when not faulty) is less CC setting than the load's current demand. That brings current limiting and output voltage drop. A form of protection.
A way to confirm is replacing with a known lighter load. A dummy load test resistor of 150 Ohm will require 100mA at 15V for example. When the CC setting is 200-250mA as recommended, there is no reason for the output voltage to drop with that test load. Else, there is likely a fault in the PSU's build or a bad part.
Most unusual will be an oscillatory situation between the PSU and some circuit as load. An oscilloscope can clearly show that and some rearrangement can be found.
Hi Salas,
I put a pair of 180Ω/1W resistors at Regs' output for testing. It was fine.
The output voltages remain +/-15V.
I am actually experimenting on powering my Whammy headphone amp with different PSU types, including UltraBiB and an opamp based regulator. Input voltage is set at +/-15V.
For R1 value, I took the stock output bias 60mA (with 10Ω resistors I am using) and add 100mA; that's how I got the 3.3Ω resistors for R1.
What should I pick for R1 if 3.3Ω is not adequate enough? 2Ω or lower?
Thank you!
I put a pair of 180Ω/1W resistors at Regs' output for testing. It was fine.
The output voltages remain +/-15V.
I am actually experimenting on powering my Whammy headphone amp with different PSU types, including UltraBiB and an opamp based regulator. Input voltage is set at +/-15V.
For R1 value, I took the stock output bias 60mA (with 10Ω resistors I am using) and add 100mA; that's how I got the 3.3Ω resistors for R1.
What should I pick for R1 if 3.3Ω is not adequate enough? 2Ω or lower?
Thank you!
It's either you will go searching by chance to meet that Whammy build's consumption with R1 substitutions, or you will measure it with a DMM set as A-meter so to plan accordingly.
In the first case 2Ω is next logical step.
It's not only about an amplifier's bias current. Think about its subsystems and its max output on worst load possible. Also possible fault if it's a first time power on test.
In the first case 2Ω is next logical step.
It's not only about an amplifier's bias current. Think about its subsystems and its max output on worst load possible. Also possible fault if it's a first time power on test.
Hello,
What values should I use for Cx, Cs, Rf and Rs?
All my boards are for 5volts output.
Someone sent me this from another forum :
"Morgan Jones goes into great detail about this topic. After 19 pages he comes to the conclusion: "We have been chasing a demon that doesn't exist(...)Ideally we would know diode capacitance, measure leakage inductance and calculate the correct resistance to set Q=0,7, but even if we don't, 1nF and 1K will almost certainly do in both high and low voltage supplies." (Linear Audio, Vol5, p. 26)"
"You need only Cs and Rs. As you might know Salas also sells the "Quasimodo" pcb (I think 2 versions, simple and more sophisticated) for testing and finding the exact values for the snubber. But as M. Jones has shown in this article it's really enough to use only RS and Cs as the measured response curves really don't change much with more precise values which one would get from calculating (not easy, also because of hard to get reliable data for the used parts) or measuring the response (very easy with Salas boards)."
So I Polaron guess that I could use 1nf for Cs and 1K for Rs.
For Rf I intend to use 0,33ohm/1W - 1W because this is for a 5V shunt.
Sorry if this post seams alitle bit confused...well it is!
What values should I use for Cx, Cs, Rf and Rs?
All my boards are for 5volts output.
Someone sent me this from another forum :
"Morgan Jones goes into great detail about this topic. After 19 pages he comes to the conclusion: "We have been chasing a demon that doesn't exist(...)Ideally we would know diode capacitance, measure leakage inductance and calculate the correct resistance to set Q=0,7, but even if we don't, 1nF and 1K will almost certainly do in both high and low voltage supplies." (Linear Audio, Vol5, p. 26)"
"You need only Cs and Rs. As you might know Salas also sells the "Quasimodo" pcb (I think 2 versions, simple and more sophisticated) for testing and finding the exact values for the snubber. But as M. Jones has shown in this article it's really enough to use only RS and Cs as the measured response curves really don't change much with more precise values which one would get from calculating (not easy, also because of hard to get reliable data for the used parts) or measuring the response (very easy with Salas boards)."
So I Polaron guess that I could use 1nf for Cs and 1K for Rs.
For Rf I intend to use 0,33ohm/1W - 1W because this is for a 5V shunt.
Sorry if this post seams alitle bit confused...well it is!
PF5102 Alternativ
Hi Salas
The small JFET PF5102 is no where to be found anymore. It was available two weeks ago, but now it is sold out or marked as "End of Life" at RS, Farnell, Digikey and Mouser (Maybe averaible @ 11 Nov!)
I have some PF5102 for the 5v reg. but is missing them for the +- regs.
Do you know a good alternative?
Would a J112 or J113 be a good alternativ: Substituting the 2N5457 – Stompville
Thanks
Hi Salas
The small JFET PF5102 is no where to be found anymore. It was available two weeks ago, but now it is sold out or marked as "End of Life" at RS, Farnell, Digikey and Mouser (Maybe averaible @ 11 Nov!)
I have some PF5102 for the 5v reg. but is missing them for the +- regs.
Do you know a good alternative?
Would a J112 or J113 be a good alternativ: Substituting the 2N5457 – Stompville
Thanks
What the Stompville link says its true. My few J113 samples are that high IDSS too.
Nonetheless, J113s of high IDSS can still work for us. That's with higher value source resistors than now so to hold their bias current back (R4 R7 R8). Up near to 1kΩ instead of 270Ω possibly. 2V drop on 1k indicates 2mA for instance which is alright for this circuit.
The difficulty is with 5V reg builds and J3 position because if you can't find low enough IDSS samples their Vgs(off) will be inevitably high. The voltage margins are of course squeezed in 5V regs. High Vgs(off) J3s will still work there but not best.
By the way, Tea has secured a large number of PF5102s he said.
Nonetheless, J113s of high IDSS can still work for us. That's with higher value source resistors than now so to hold their bias current back (R4 R7 R8). Up near to 1kΩ instead of 270Ω possibly. 2V drop on 1k indicates 2mA for instance which is alright for this circuit.
The difficulty is with 5V reg builds and J3 position because if you can't find low enough IDSS samples their Vgs(off) will be inevitably high. The voltage margins are of course squeezed in 5V regs. High Vgs(off) J3s will still work there but not best.
By the way, Tea has secured a large number of PF5102s he said.
Thanks
There are also several shops on eBay selling the PF5102. Some of the shops are probably fine but I am not sure which one are selling 100% genuine parts...
To clarify you said 2mA will 'work fine'. Is this the value with the specified JFET?
I will measure few J113s with some degeneration resistor values tomorrow to tell you which one value brings them to an agreeable average Id for this circuit.
Salas, I know this is out of topic, but I am asking this, since we're talking about J113s in this moment.
I am desperate to find 6 Toshiba 2SJ313's matched for Vgs at 10 mA.
This is for an Aleph III that I am finishing - 3 per board.
Does anyone knows where to buy these Jfets, matched like this?
I can make an exchange. Psu, dec, caps...or buy it.
Thank you!!
I am desperate to find 6 Toshiba 2SJ313's matched for Vgs at 10 mA.
This is for an Aleph III that I am finishing - 3 per board.
Does anyone knows where to buy these Jfets, matched like this?
I can make an exchange. Psu, dec, caps...or buy it.
Thank you!!
Back to J113. I run some drain current tests with various resistors and various voltages today. Results suggest:
For 5-10mA IDSS (less likely to find in current production J113) keep the same 270 Ohm R4 R7 R8 value. For 10-20mA IDSS use 560 Ohm R4 R7 R8. For >20mA IDSS use 750 Ohm R4 R7 R8.
When having few samples from different IDSS ranges, use the appropriate resistor value to each sample's position. J1 corresponds to R4, J2 to R7, J3 to R8. Prefer the lowest IDSS samples for J3 when its for 5V rail set regs.
Sorting J113s for IDSS takes a semiconductors tester or to put them on a breadboard and connect G+S pins together. Use +15V with a milliamper meter between +V and D (drain) pin in series. 0V to S pin. That is the datasheet's test voltage. Using a 9V battery will give bit less IDSS but still indicative.
Test the samples replacing one by one in the rig. Measure carefully to avoid accidental shorts that can break the mA scale DMM fuse. Set 50mA output current limit if you use a bench supply so not to worry about shorts (CC function). Note their Idss results on paper. Place each sample next to its mA note.
Because J113 is 35V rated don't use higher than 28VAC transformer. Thus the Ubib's suggested max output shortens to 33VDC for good enough Vin-Vout margin when using this JFET type.
Although PF5102 and J113 are brothers, in general I prefer the PF5102 due to its lower more predictable IDSS range and 5V higher official Vmax spec. Seems like that type is a subset of J113 production with more controlled characteristics. While coming from the same chopper JFETs family it is alternatively described as low frequency low noise amplifier in the manufacturer's website.
P.S. I will refer this information in post #1 also
For 5-10mA IDSS (less likely to find in current production J113) keep the same 270 Ohm R4 R7 R8 value. For 10-20mA IDSS use 560 Ohm R4 R7 R8. For >20mA IDSS use 750 Ohm R4 R7 R8.
When having few samples from different IDSS ranges, use the appropriate resistor value to each sample's position. J1 corresponds to R4, J2 to R7, J3 to R8. Prefer the lowest IDSS samples for J3 when its for 5V rail set regs.
Sorting J113s for IDSS takes a semiconductors tester or to put them on a breadboard and connect G+S pins together. Use +15V with a milliamper meter between +V and D (drain) pin in series. 0V to S pin. That is the datasheet's test voltage. Using a 9V battery will give bit less IDSS but still indicative.
Test the samples replacing one by one in the rig. Measure carefully to avoid accidental shorts that can break the mA scale DMM fuse. Set 50mA output current limit if you use a bench supply so not to worry about shorts (CC function). Note their Idss results on paper. Place each sample next to its mA note.
Because J113 is 35V rated don't use higher than 28VAC transformer. Thus the Ubib's suggested max output shortens to 33VDC for good enough Vin-Vout margin when using this JFET type.
Although PF5102 and J113 are brothers, in general I prefer the PF5102 due to its lower more predictable IDSS range and 5V higher official Vmax spec. Seems like that type is a subset of J113 production with more controlled characteristics. While coming from the same chopper JFETs family it is alternatively described as low frequency low noise amplifier in the manufacturer's website.
P.S. I will refer this information in post #1 also