a different phono stage

[johnrtd]

As some of you know we (Hawk Audio) offer a phono pre-amp with tubes and an op amp at the input. The latter to have a low noise ane wide band possibility for MC cartridges.
We sold only a few kits of this design, probably because of the complexity and the costs (double power supply!). The current users are very happy though.

Thinking it over I considered designing a "solid state" circuit. Nowadays there are some very nice op amps available, specially designed for high end audio purposes. So, in principle, it is possible to create a relatively cheap design this way.

A problem with IC's is that they, as any audio circuit, are susceptible to variations in the supply. More so at higher frequencies (not mentioned in the specifications). So one should take measures to be sure the supply does not deteriorate the sound quality.

The easy way is using integrated stabilizers. The simple ones such as the 78.. and 79.. types have a disadvantage while having a rather large capacitance between input and output. This way noises and spikes from the mains easily enter the audio circuit. The LM317/337 are better ones, but more complex and some extra components are needed to have them functioning the right way.

A different thing is the "interaction" of the different amplifying stages via the supply connections. Best thing is to isolate those circuits from each other. So a second stabilizer is needed, one for every stage in the circuit. We have some experience with shunt regulators. Those have the advantage of a large series resistor giving an extra isolation from the common supply. We often see integrated regulators, such as the TL431 type, used in circuits. But those have a rising impedance when the frequency goes up. So a discrete regulator is a better choice (faster!) although it's more complicated and more parts are needed.

Have a look at the block drawing of the main supply:


Looking from the left there are two resistors between the mains transformer and the bridge rectifier. This is standard in all our supplies (also in tube circuits!). Those resistors prevent spike-like signals originating from the rectifiers. Thereafter there's a capacitor, a stabilizer and again a capacitor. The value of the capacitors is relatively low, but they have superior specifications: low esr, 105°C and long life.
The voltage goes down from 40 Volts toward 24 Volts, so the regulator has sufficient headroom to fulfill it's task.

Next is the block drawing of the amplifier:


At the input three switches are chosen in order to let the input and amplification factor be sufficient for a variety of cartridges.

The first stage is a very special (bipolar!) op-amp having current mirrors designed as amplifiers. The shunt regulators stabilize a voltage of + and -12,6 Volt, enough to give the circuit ample headroom with any type of input signal.

The second and third stage are together in a double op-amp. This one is a brand new type from Analog Devices. The supply is stabilized at + and -17 Volt. Inbetween is the passive RIAA network.

[johnrtd]

As stated before the power supply is important. It should prevent noises from the mains or hum enter into the amplifying circuits.
We opted for a classical supply (not an SMPS) as a switching supply generates a large (HF) noise level. Although it must be said that things are developing rapidly and I expect that in the future such a switching supply might become an option at a reasonable price. The current classical circuit is this one:

It starts with a toroid transformer. This has a rather small "field" around it so it can be placed nearby the amplifier circuitry.
The secundary windings are 24 or 30 Volt which gives 36 - 45 Volt DC after the rectifier.
Between the transformer and the bridge rectifier some small 10 Ohm resistors are placed suppressing the generation of spiky signals by the silicon rectifier. This also has to do with the properties of the capacitors (C11 and C13) used for the filtering. In the prototype we placed the beautifully specified Panasonic ones out of the FC-series. The value is only 1000 µF - 50 Volt because the total current drawn by the circuit is rather small, some 250 mA. With such a small value the price is acceptable.
The stabilizers are LM317/337 types. Those bring the voltage down to + and -24 Volt. The Wima capacitors C12 and C14 take care of the HF behaviour of the LM types.

Let's look at the amplifying circuits:

There are two op amp circuits, each half of the OPA2228. That op amp is specified as having a noise level of 3 nV/VHz a bandwidth of 33 MHz and a slew rate of 10V/µsec. Both op amps amplify 20 dB so with 3 mV at the input there would be 300 mV at the output.
Inbetween both op amps there's the RIAA circuit (R19, 20, C6, 7, 8). The capacitors are 2% polystyrene types and this should give a deviation from the RIAA curve of maximum 0,3 dB.
On top there's the shunt regulator with TR9, 10 and Z3. The value of R29 determines the current as we have 24 Volt at the input and 17 Volt at the output. So with a 7 Volt difference and a 1 kOhm resistor there would be a current of 7 mA. Both op amps use some 3,5 mA which totals to 7 mA. We want the shunt to use at least the same current, which totals to 14 mA. A practical value for that resistor now would be 475 Ohm.
C10, connected directly with the power supply, is a nice Panasonic again. The capacitor at the output of the shunt is an Os-Con type which performs good at higher frequencies.
Zener diodes generate some noise so we added a small Wima capacitor (C17, 18)
Each circuit, left and right channel, has it's own shunt regulators so there will be very little cross talk between the channels.

(to be continued)

[johnrtd]

When busy on the design I discussed it with my colleague Jacco Dekkers. Jacco is an engineer who finished his studies with a SET amplifier design at our labs (this JD-10 is still downloadable).
He suggested a different solution for the power supply:



When looking at it the input, connected to the bridge rectifier, is on the left side. There's a 7824 as a first stage stabilizer. Then two discrete series regulators follow each with BC/BD types of transistors. Each amplifier stage is fed with an integrated current source.
Jacco stated that when simulating this circuit the noise at the output is around 3 nV!
Probably because I'm old fashioned I don't "believe" simulations. So we'll wait and see after Jacco has built his circuit what "real world" figures can be measured.
For this moment we'll keep using the shunt regulators although it must be said that Jacco's series regulators consume far less.
Any comments or suggestions?

(to be continued)

[anthonyTD]

When busy on the design I discussed it with my colleague Jacco Dekkers. Jacco is an engineer who finished his studies with a SET amplifier design at our labs (this JD-10 is still downloadable).
He suggested a different solution for the power supply:

Cascode Power Supply

A pity, but I cannot copy this drawing on this board.
When looking at it the input, connected to the bridge rectifier, is on the left side. There's a 7824 as a first stage stabilizer. Then two discrete series regulators follow each with BC/BD types of transistors. Each amplifier stage is fed with an integrated current source.
Jacco stated that when simulating this circuit the noise at the output is around 3 nV!
Probably because I'm old fashioned I don't "believe" simulations. So we'll wait and see after Jacco has built his circuit what "real world" figures can be measured.
For this moment we'll keep using the shunt regulators although it must be said that Jacco's series regulators consume far less.
Any comments or suggestions?

(to be continued)

hi john,
looking at the circuit briefly i can see one main problem, or rather something that i would be dead against doing having experienced what affect it has on sonics, that is, the resistors you have in line before the bridge rectifiers, as you say, they will to a point suppress some noise, but will also cause a reduction in percieved dynamics, and bass grip, after-all what you are realy doing is raising the power supply impedence, which is something that in my opinion should be avoided at all costs.
regards, anthony.

[johnrtd]

Dear Anthony

Mind you the following was published formerly on this site! Also you could have a look at our tips.

When developing a SET amp around 1994 a student of mine remarked that he saw some "funny spikes" when looking at the output of the rectifier and with the oscilloscope switched to a high sweep frequency. I looked into it and came up with two solutions: an extra cap in parallel with the big electrolytic one or a resistor in advance of the rectifier.
We found that the spiky misery was also found in the amplifier circuitry (without fb).
I oppose to paralleling capacitors in the supply. We did this in the eighties, partly because the electrolytic capacitors then had a rising ESR from 100 Hz and up. That time is past, we now have much nicer caps. And .......... when filtering the supply with caps of different values in parallel there might rise a phase problem.
So we tried out some small resistors and the audible results were very good! Thereafter we used these spike-stop resistors in all our amplifier designs, be it class-A, -B or -D. And, of course, also in our pre amps.
A few years later we found that the Philips company uses this technique since the early seventies in all their medical instruments! So there's nothing new.
Being an editor of an audio magazine at that time we also had the possibility to try this on a variety of amplifiers from various manufacturers. In all cases we had good results.
After publishing this "trick" a lot of audio amateurs tried it with good results, even some guys around the "Art of Sound Forum".
Regarding the loss of dynamics one should realize that the main source for instantaneous power is in the big capacitors. These are loaded from the transformer of course, but then we have to do with the internal resistance of the transformer + rectifier. Using a small value resistor the loss of dynamics is quite small (in power amps we use 0,1 Ohm!). In preamps such as the one in the article the current from the supply is quite steady so no dynamics whatsoever are involved.

BTW when using "fast recovery" diodes or rectifiers the resistors are not sufficient. Then each diode should be paralleled with a small capacitor, preferably a polycarbonate one.

John

[anthonyTD]

Dear Anthony

Mind you the following was published formerly on this site! Also you could have a look at our tips.

When developing a SET amp around 1994 a student of mine remarked that he saw some "funny spikes" when looking at the output of the rectifier and with the oscilloscope switched to a high sweep frequency. I looked into it and came up with two solutions: an extra cap in parallel with the big electrolytic one or a resistor in advance of the rectifier.
We found that the spiky misery was also found in the amplifier circuitry (without fb).
I oppose to paralleling capacitors in the supply. We did this in the eighties, partly because the electrolytic capacitors then had a rising ESR from 100 Hz and up. That time is past, we now have much nicer caps. And .......... when filtering the supply with caps of different values in parallel there might rise a phase problem.
So we tried out some small resistors and the audible results were very good! Thereafter we used these spike-stop resistors in all our amplifier designs, be it class-A, -B or -D. And, of course, also in our pre amps.
A few years later we found that the Philips company uses this technique since the early seventies in all their medical instruments! So there's nothing new.
Being an editor of an audio magazine at that time we also had the possibility to try this on a variety of amplifiers from various manufacturers. In all cases we had good results.
After publishing this "trick" a lot of audio amateurs tried it with good results, even some guys around the "Art of Sound Forum".
Regarding the loss of dynamics one should realize that the main source for instantaneous power is in the big capacitors. These are loaded from the transformer of course, but then we have to do with the internal resistance of the transformer + rectifier. Using a small value resistor the loss of dynamics is quite small (in power amps we use 0,1 Ohm!). In preamps such as the one in the article the current from the supply is quite steady so no dynamics whatsoever are involved.

BTW when using "fast recovery" diodes or rectifiers the resistors are not sufficient. Then each diode should be paralleled with a small capacitor, preferably a polycarbonate one.

John

hi john,
i agree that this solution works as far as suppresing the spikes/noise, but going by the experience i gained while designing my own pre-amp [soul-mate] i could definately hear the diffrence with reletively small resistors in line with the 100va toroids i was using in each channel.
however, the caps in parralell with the diodes is a very worth while move indeed.
anthony.

[Mike]

Interesting stuff regarding the diodes and parallel caps.

How much do you guys thing it would help when using schottky diodes? What type/value caps would you recommend?

Cheers,
Mike.

[johnrtd]

So Anthony I understand it's a pity for the people working in a medical environment, having to control heart failures for instance, that the Philips equipment suffers from dynamics but be sure that noise (a big problem in medical equipment) is sufficiently suppressed. And of course all my customers suffer. A pity again.

John

[johnrtd]

Interesting stuff regarding the diodes and parallel caps.

How much do you guys thing it would help when using schottky diodes? What type/value caps would you recommend?

Cheers,
Mike.

Hi Mike! I'm sorry I don't "thing"! In tube amps I suggest to use 0,01 µF - 1000 Volt MKC types from Wima. In solid state (transistor) equipment 0,01 µF - 400 Volt will do. And better use "slow" Schottky!

John

[Mike]

Oops!... another quality typo!

Thanks John.

P.S. 'Slow' schottkys??? huh?