John Curl's Blowtorch preamplifier

[1audio]

Originally posted by KSTR
This represents a "chopped" total leakage capacitance that might further "chop up" GND currents to/from other gear. With the simple bridge the coupling capacitance is always in place and doesn't change that much (but still to some extent), which might give way more benign GND-current spectra.

- Klaus

These currents are much more significant than they might first seem to be. These currents end up going through all the grounds in the system. On the low level stuff they can subtly modulate the audio. And the "HF chop" (actually the charging current for the filters) can have a broad spectrum, and the magnetic field can do the same. I currently float supplies and derive ground at the regulator to keep it clean.

[MikeBettinger]

Originally posted by john curl
I also use the dual bridge diode array, also. The A mod for the Vendetta Research phono stage was just this. Later, I also upgraded to high speed, soft recovery diodes, and this was called the D mod. It is not a universal fix however, since the +/- loads must be matched.

Was the result of the loads not being matched a higher 120hz noise floor?

Mike.

[john curl]

No, DC shift.

[MikeBettinger]

Originally posted by john curl
No, DC shift.

Didn't the regulators compensate for the DC shift? Or was this a dynamic shift of some sort?

The only reason I asked is that I have trouble getting a solid "minimal hum" noise floor with the dual bridge approach and I'm curious if there is something I'm missing and this is not an issue.

Mike.

[wrenchone]

As a bit of historical perspective, Motorola published a short applications note on synchronous rectification using germanium transistors - guess how old that is...

[MikeBettinger]

Originally posted by MikeBettinger
Didn't the regulators compensate for the DC shift? Or was this a dynamic shift of some sort?
Mike.

I thought about your comment, as usual I'm in another universe.

I would just trim one of the regulators to remove the offset. I experience difficulty (relative to a single bridge an a solid reference to the centertap) with the dual bridge and virtual ground.

As Gilda would say, nevermind.

Mike.

[KBK]

Originally posted by 1audio

KBK:
If I didn't know better I would describe your note as first order audiophile gobbledygook. But I do know a little better. I have had no success with those tricks myself but I suspect they may help, I'm not sure how. Does the pot material, wiper allow etc. make a difference with this trick? I abandoned the pots in premium products a long time ago and used Tech Labs switches with Vishay resistors. Nothing else I tried came close. I went on to look at more exotic things like airtight enclosures of non-magnetic non-conductive materials and vibration damping tricks. [/B]

I believe you are correct in fixing the issue at the source instead of chasing the closing of little dirty and resonant barn doors. However, properly handled it has the capacity to give a notably better quality to a lower priced variable wiper type control unit.

[john curl]

The DC shift is due to the lack of a DC ground return.

[Bobken]

Originally posted by 1audio

That is the pot we used in the production Spectral DMC-10. It had the best sound of the practical options at the time.

Hi Demian,

Apologies to John for a slight OT excursion, but I have only just seen this, and these comments do relate to HQ preamp circuit design choices.

My DMC-10 was fitted from new with the 10k Alps Blue (or is it Black-Beauty?) pot you mentioned and (Stackpole) switched discrete resistors for the balance control. After returning it to Spectral following a minor fault, when I also agreed with Rick Fryer to having the latest 'Delta Revision' update carried out, it was returned with a Noble pot, instead, and a 'detented' 100k Noble rotary pot for balance purposes.

Unfortunately, I was unable to assess precisely what this did, in isolation, for the sonic results. Several other significant mods were carried at the same time including the addition of cascoded dual J-Fets (on piggy-back daughter-boards) in the front end, together with some changes to the HF stabilising arrangements, to mention a couple.

Shortly afterwards, I changed just the Noble volume pot to a (wickedly costly) hand-selected conductive-plastic Penny & Giles pot, and like you, I still found this to be 'wanting' in performance, although it was an improvement on the Noble pot it replaced.

I was only finally satisfied when I replaced the entire volume/balance arrangements with 2 Shallco switches populated with all Vishay bulk-foils. This meant operating 2 controls to set the volume, although being 'stepped' - accurate L/R balance was still easy, but the improvements in sonic performance all around were so so marked that this was a very minor inconvenience.

This would have been over 20 yrs ago IIRC, but since then I would never use any proprietary rotary pots in any of my designs, although for a recent commercial commission I did adopt relays, purely for 'enforced' remote control purposes.

Here I concur with Sigurd, and have found that the TX2352 (naked Vishay resistors) from Texas Components are truly excellent, even when switched through relays (although Shallco 'manual' switching is sonically superior!) and, of course, they can be made in tight tolerances to ANY value to 6 decimal places, which provides absolute accuracy to fractions of a dB of attenuation.

Also, to reinforce John's earlier comments on the audibility of ferrite beads (which I will not use anywhere nowadays following many years of careful listening trials - although initially I thought that the potential problem was restricted merely to high-current locations), I also removed the ferrite beads in the phono inputs at one stage, and there was a clear sonic benefit subjectively. IIRC, these were also added during the update revision rather than being present from when the DMC-10 was new, but they were detrimental to the sonic results, as I duly discovered.

Incidentally, what was the true purpose of the 400R resistors located directly between the inputs and outputs of the regulators on the DMC-10 PCB? Intuitively, I never came to terms with their inclusion, although clearly you must have a had a good reason for this addition which is in effect a resistive 'bypass' to the regulators, themselves.

Regards,

[jan.didden]

Originally posted by KSTR
I also like this arrangement pretty much, as it allows for various forms of symmetry and separation, use of lower diode breakdown voltages etc.

There might be one noise penalty, though... (I haven't measured this in detail so far, so it's only armchair reasoning) :
When the diodes don't conduct, most of the cycle in fact, the xformer leakage capacitance is seriesed with the diodes' capacitances, while when the diodes conduct the xformer is connected directly. This represents a "chopped" total leakage capacitance that might further "chop up" GND currents to/from other gear. With the simple bridge the coupling capacitance is always in place and doesn't change that much (but still to some extent), which might give way more benign GND-current spectra.

- Klaus

I have been using discrete bridges since the early
90's build up from fast but soft recovery diodes. At the time, XYIS (sp?) had some that were relatively expensive. In my last amp I used Philips BYV32E-200, which from the spec sheet seem a huge overkill, but they are less than a buck each. They are actually two diodes in a plasic TO3 but I just parallel them on the board.

They are so 'soft' recovery that you can skip the snubber/cap across them which of course eliminates another path from mains to your circuit.

Jan Didden