I don't know enough to design a ground plane but I'll try to explain using what tools I have.
R15 is fed from the +ve and goes to Q20 and thence to the ground plane connected emitter.
There are a lot of capacitors connecting the ground plane back to the +ve.
One is a small high frequency capable version (C14), all the others appear to be electrolytics.
Any high frequency interference current passing from +ve via R15 to Q20 across that slot in the plane has to find a low impedance route via C14 back to +ve. The lowest impedance route for HF is parallel to the Flow route.
Does your ground plane and placement of components allow the current to find a low impedance route for the high frequency interference?
Maybe what little extra impedance resulting from crossing over that slot is of no consequence, but I can't make that assessment, because I can't design planes. I can't assess whether the location of C14 helps or hinders interference passing into the plane.
You need to make those assessments. You are the designer implementing the plane.
The general rule I see being put forward by ground plane designers is that: no component, nor trace should pass across a cut/slot/gap in the current returning plane. H.Ott has sketches showing what we should NOT implement.
As I have said already, when I saw the (pink area) layout of post 3529 I thought about this general rule. Slots everywhere and that prompted my question.
R15 is fed from the +ve and goes to Q20 and thence to the ground plane connected emitter.
There are a lot of capacitors connecting the ground plane back to the +ve.
One is a small high frequency capable version (C14), all the others appear to be electrolytics.
Any high frequency interference current passing from +ve via R15 to Q20 across that slot in the plane has to find a low impedance route via C14 back to +ve. The lowest impedance route for HF is parallel to the Flow route.
Does your ground plane and placement of components allow the current to find a low impedance route for the high frequency interference?
Maybe what little extra impedance resulting from crossing over that slot is of no consequence, but I can't make that assessment, because I can't design planes. I can't assess whether the location of C14 helps or hinders interference passing into the plane.
You need to make those assessments. You are the designer implementing the plane.
The general rule I see being put forward by ground plane designers is that: no component, nor trace should pass across a cut/slot/gap in the current returning plane. H.Ott has sketches showing what we should NOT implement.
As I have said already, when I saw the (pink area) layout of post 3529 I thought about this general rule. Slots everywhere and that prompted my question.
after you posted
Last edited:
If I understand you correctly the path you are concerned about is Q20 (COM/GND) to R15 (V+) via C14, illustrated below. And - again I'm not 100% sure I'm seeing what you are seeing - your concern isn't so much the distance of C14 from those components as it is the short section of top trace connecting R13-Q17 which breaks the ground plane in that region.
I'm aware that it's said to be a bad idea to artificially gate a ground plane into sections, though I've also seen this recommended for separating analog and digital grounds of DAC circuits for example... we could discuss the relative merits but I think the main point here is that the VSPS circuit has a bandwidth well below the kinds of frequencies where these kinds of issues become a significant concern. The ground plane is a just a big, flat wire connecting the returns together. Perforating it with holes and even a few slots here and there does not appreciably alter its DC resistance, or it's impedance at the frequencies of interest. As long as the components connecting to it do so in the right order such that power supply and output current do not impinge on the input signal currents, everything else will take care of itself.
For me, the main worry for the VSPS layout is avoiding the classic mistake of a ground loop condition where the return load currents are impressed onto the noninverting input return. A secondary worry is power supply currents getting into the either op amp inputs. Third on the list of layout considerations is having short distances from the op amp bypass capacitors to the op amp supply pins. The latter should take care of any residual noise on V+, such as the interference signal you posited in R15/Q20.
I'm aware that it's said to be a bad idea to artificially gate a ground plane into sections, though I've also seen this recommended for separating analog and digital grounds of DAC circuits for example... we could discuss the relative merits but I think the main point here is that the VSPS circuit has a bandwidth well below the kinds of frequencies where these kinds of issues become a significant concern. The ground plane is a just a big, flat wire connecting the returns together. Perforating it with holes and even a few slots here and there does not appreciably alter its DC resistance, or it's impedance at the frequencies of interest. As long as the components connecting to it do so in the right order such that power supply and output current do not impinge on the input signal currents, everything else will take care of itself.
For me, the main worry for the VSPS layout is avoiding the classic mistake of a ground loop condition where the return load currents are impressed onto the noninverting input return. A secondary worry is power supply currents getting into the either op amp inputs. Third on the list of layout considerations is having short distances from the op amp bypass capacitors to the op amp supply pins. The latter should take care of any residual noise on V+, such as the interference signal you posited in R15/Q20.
Attachments
S-Reg tested
I have a standalone version of the S-Reg used in the phonoclone 4, so today I put one together - mostly to make sure the trimmer provided a reasonable sensitivity for adjusting the output. I built it with R1,2 22.1 ohms, and the source current was about 26 mA as expected. The load resistor draws 10 mA.
Yes, it works just fine. With BD135/136 transistors and 1k trimmer, the output was 11.5 V and the trimmer easily get the V+ and V- matched to within 25 mV. Once set, the voltages remained within that tolerance.
The scope photo shows the V++ input ripple (65 mV rms) and V+ output ripple (~1 mV, that trace displayed at 5 mV/cm). Whatever residual signal is seen on the output is probably noise pickup.
I have a standalone version of the S-Reg used in the phonoclone 4, so today I put one together - mostly to make sure the trimmer provided a reasonable sensitivity for adjusting the output. I built it with R1,2 22.1 ohms, and the source current was about 26 mA as expected. The load resistor draws 10 mA.
Yes, it works just fine. With BD135/136 transistors and 1k trimmer, the output was 11.5 V and the trimmer easily get the V+ and V- matched to within 25 mV. Once set, the voltages remained within that tolerance.
The scope photo shows the V++ input ripple (65 mV rms) and V+ output ripple (~1 mV, that trace displayed at 5 mV/cm). Whatever residual signal is seen on the output is probably noise pickup.
Attachments
Last edited:
S-reg have temperature drift..
That's not important for opamp performance, but could become if you eliminate C3..
@dimkasta
The output voltage varies with the Vbe of the transistors, which gets smaller as they heat up. There's a few minutes of stabilization on start up where the output slides 30 mV, but it is stable afterwards and positive and negative rails move in lockstep. No, as vulejov correctly notes this cannot be tweaked to fiddle with the op amp output offset with any kind of reliable result.
@vulejov
re BC/BD It's a bit borderline I admit. What do you think the maximum power should be for a BC337/327? Datasheet says 625 mW and in normal operation the S-reg transistors are under 300 mW.
I could reduce the shunt current in the VSPS since it only uses one opamp, but my thinking at the moment is just convert the layout to use BD135/136 like the standalone board.
The output voltage varies with the Vbe of the transistors, which gets smaller as they heat up. There's a few minutes of stabilization on start up where the output slides 30 mV, but it is stable afterwards and positive and negative rails move in lockstep. No, as vulejov correctly notes this cannot be tweaked to fiddle with the op amp output offset with any kind of reliable result.
@vulejov
re BC/BD It's a bit borderline I admit. What do you think the maximum power should be for a BC337/327? Datasheet says 625 mW and in normal operation the S-reg transistors are under 300 mW.
I could reduce the shunt current in the VSPS since it only uses one opamp, but my thinking at the moment is just convert the layout to use BD135/136 like the standalone board.
For example we could reduce the shunt current by changing the resistors to something like the following. Max transistor dissipation is ~150 mW.
According to LTSpice the op amps draw less than 3 mA each, but same allowance must be made for manufacturer variation and model substitution.
According to LTSpice the op amps draw less than 3 mA each, but same allowance must be made for manufacturer variation and model substitution.
Attachments
I think that would be no problem to find BD139/140 with hfe similar to BC327/337..
About listening comparasion..
I use old VSPS board with LM317/337 PSU as MC phono, AD797 with 60-65dB gain.. S-reg have better perfomance than few decades old voltage regulators, but I'm note sure that we can hear difference.. maybe I try with final VSPS 4 version..
Empirically, changing regulators seems to make about as much difference to the sound as changing op amps. Small but significant enough for people to develop a preference.
I don't want to pretend this is about increasing measurable "performance", as from a standpoint of numbers either a simple Z-reg or your LM317 is sufficient. The PSRR of 100dB+ of the op amp ICs makes this so. I just want to try a shunt regulator this time around, as being something of the opposite design strategy to the X-reg in the phonoclone 3.
I don't want to pretend this is about increasing measurable "performance", as from a standpoint of numbers either a simple Z-reg or your LM317 is sufficient. The PSRR of 100dB+ of the op amp ICs makes this so. I just want to try a shunt regulator this time around, as being something of the opposite design strategy to the X-reg in the phonoclone 3.
Hi all,
I had a go at building Richards phono stage clone, its a bit rough as I was using cases from a previous 'failed' project.
I'm keeping it as simple as possible at the moment - using an already existing external power supply - may ditch that it the future and populate the x-reg!!
No hum, as long as I keep any transformers away, and I've found that using no R1 resistor, ie just hard wired, with around 600-800 Ohm for R2 works best with my Dynavector DV17 MkII into my Naim pre.
Still experimenting, the Dynavector has a Internal Dc R quoted as 38 Ohms and outputs 0.23 mv.
Thanks to Richard for a great kit and a great service!
Just a little update to say that (perhaps not surprisingly
) I have had better results by first ditching the external 'interesting! power supply and populating the x-reg on the phonoclone 3.Much better in fact - mid and low end cloudiness eradicated and very happy to dig out 'most' of my old albums!
Attachments
Nice timing, that fits in very nicely with the "how does voltage regulation affect the sound" discussion above. No reason to imagine the regulator module you were using before was failing the phonoclone circuit in any way, but here is another example where the difference is enough to find a preference.
My own opinion is a lot of it is where the regulator is as much as what it is (or how much). I've found the best location is as close to the op amps as possible.
My own opinion is a lot of it is where the regulator is as much as what it is (or how much). I've found the best location is as close to the op amps as possible.
hi RJM,
Do you have plans to release Phonoclone 4.0b soon ? I've never built Phonoclone and have a strong interest to build one to compare with head-head on my Pearl2 build.
If you have plan to release a 4.0b version soon, i'll wait or else I will go ahead and order 3.0 version from you site.
Thanks,
Tom
Thank you so much for the fast and prompt reply Richard. In this case, i will wait for a couple of more weeks till the newer version comes out! Please do let us know when the new V4.0b board arrive.
Out of curiosity, do you know between S-reg and X-reg, which one will have better sounding result when pairing with op-amp based phono stage ? I'm thinking of buying a separate set of either S-reg or X-reg to replace and compare with my current LM337 volt regulator ckt in my current PSU setup to a couple of phono preamps that I built in the past. I have built both jfet based and opamp based phono preamp in the past.
Thanks,
Tom