I have a very strange problem indeed.
I built Kevin Gilmore's CFA3 amplifier, the I added Kevin Gilmore's GRLV linear supply to power it.
When I power the amp from a cold start, a high-frequency, almost whistle-like sound shows up on the output. It las a few seconds (longer if it's cold, less if it's hot) and then goes away. While that sound is heard, I can make it go away by touching the 4-pin XLR output socket.
Once the amp has been running a while, I can power it off, wait a couple of minutes, and power it back up with no noise of any kind.
I think this strange noise didn't come with the previous power supply.
So, quite a few clues, but none of it makes sense to me.
Any wizard can school me on that one?
I suspected oscillation, transformer ringing, and various other things, but I can't convince myself they could be valid reasons.
I built Kevin Gilmore's CFA3 amplifier, the I added Kevin Gilmore's GRLV linear supply to power it.
When I power the amp from a cold start, a high-frequency, almost whistle-like sound shows up on the output. It las a few seconds (longer if it's cold, less if it's hot) and then goes away. While that sound is heard, I can make it go away by touching the 4-pin XLR output socket.
Once the amp has been running a while, I can power it off, wait a couple of minutes, and power it back up with no noise of any kind.
I think this strange noise didn't come with the previous power supply.
So, quite a few clues, but none of it makes sense to me.
Any wizard can school me on that one?
I suspected oscillation, transformer ringing, and various other things, but I can't convince myself they could be valid reasons.
I have a DAC connected to it with 2 X 3-pin XLR cables. There’s only 1 output via the 4-pin XLR socket.
The output is not direct, it goes to a protection board, which adds a delay on start-up and disconnects the headphones if more than a certain DC offset is detected.
There’s no Zobel network on the output.
The output is not direct, it goes to a protection board, which adds a delay on start-up and disconnects the headphones if more than a certain DC offset is detected.
There’s no Zobel network on the output.
Could it be (one of) the regulators' instability? Unlike the CFA3, GRLV packs quite some feedback with rather, er, minimal frequency compensation. Regulators of that kind a known to become unstable after trivial changes such as replacing electrolytic capacitors. Check supply rails with a scope when switching it on.
Since, it's not OK, I might as well revisit my grounding scheme for this build.
Here are 2 amazing pictures I made with my own hands. (I am the Picasso of wiring diagrams, am I not?
It's not quite as ordinary as some people will want.
So, I need to explain some things.
1) It's a 2-chassis set-up.
2) At the DAC, pin 1 of the 3-pin XLRs connects to earth, and I connect pin 1 to the chassis of the amp box at inlet. So, both PSU and amp chassis are earthed at a common point via the power strip both the PSU and the DAC are connected to.
3) Inside the amp box, I use shielded cabling between 3-pin XLRs and amp board inputs. The shield is connected to pin 1 on the inlet side only.
4) Inside the amp box, I used shielded cabling between the amp board outputs and the the protector board. The shield is connected to circuit ground on the amp boards' side only.
5) The cable between protector board and 4-pin XLR is not shielded, but very short.
Here are 2 amazing pictures I made with my own hands. (I am the Picasso of wiring diagrams, am I not?
It's not quite as ordinary as some people will want.
So, I need to explain some things.
1) It's a 2-chassis set-up.
2) At the DAC, pin 1 of the 3-pin XLRs connects to earth, and I connect pin 1 to the chassis of the amp box at inlet. So, both PSU and amp chassis are earthed at a common point via the power strip both the PSU and the DAC are connected to.
3) Inside the amp box, I use shielded cabling between 3-pin XLRs and amp board inputs. The shield is connected to pin 1 on the inlet side only.
4) Inside the amp box, I used shielded cabling between the amp board outputs and the the protector board. The shield is connected to circuit ground on the amp boards' side only.
5) The cable between protector board and 4-pin XLR is not shielded, but very short.
Thanks for your input, I truly appreciate it.
Alas, I have contractors doing work near where my bench is, so I am left with my imagination only right now. But I'll explore this later.
I think it was rather stable when I tested the supply boards after building them, but that was without load. (I was under the assumption this design was proven stable through the experience of many builders.)
Well, more than 2 dozens of cold starts later and the problem has not returned.
I guess an impedance mismatch between the DAC and the amp caused instability at startup and the addition of the preamp in between eliminated the mismatch.
As for my grounding scheme, it relies on the incoming pin 1s to be grounded on the preamp side, which is the case here and the noise floor is very low.
So, I decided to leave that alone.
I guess an impedance mismatch between the DAC and the amp caused instability at startup and the addition of the preamp in between eliminated the mismatch.
As for my grounding scheme, it relies on the incoming pin 1s to be grounded on the preamp side, which is the case here and the noise floor is very low.
So, I decided to leave that alone.