The typical advantages of tubes, which can be achieved in any reasonable design, according to me are
These good qualities can be found even in tube power amps, but here the task is more critical and requires an output transformer whose design (and especially whose testing) is rather out of the reach of a normal audiophile (each prototype of an output transformer costs a lot of money...).
As a matter of fact there are lots of problems even in low power design: microphony, noise (very often even fine "musical" tubes have big noise and microphony problems), AC coupling, often not so good bass control, lack of transparency and missing neutrality and so on. One major issue is high voltage, which is for sure a kind of "moral" problem for anyone suggesting such a design for DIY.
By the way there are a lot of commercial solid state design that sound rather good, so it is definitely not so sure that solid state cannot achieve the same quality results as a tube design.
The point, according to me, is simply that tube and solid state design have normally different qualities, and I am afraid that this is not due only to technical issues and basic features of the devices, but also to a different design philosophy between solid state and tube designers.
I have listened to some amps in which the "old tube sound" was especially looked for. One example: at last Top Audio Show in Milan I found a tube amp manufactured by a little Italian company which is crowded with paper and oil power supply capacitors, but uses low cost film coupling capacitors. I asked the designer, and he told me that he was looking for that kind of sound. Quite coloured indeed and lacking transparency, by the way: they were among the worst capacitors I tested from this point of view...
Actually solid state design normally has a far better bass control and a good musicality, even though far different from tubes' one: this is more connected to rhythm and impact, while tubes' one is more concerned with a certain levity of the sound, an intrinsic elegance, a light warm old-time dance effect (Vienna waltz I mean, not techno...).
So in the end the choice between solid state and tubes is rather a matter of personal taste than anything else.
In DIY anyway, at least here in Italy, solid state is nearly off limits; discreet components design is felt to be totally out of fashion, while op amp design is really considered no longer acceptable, given all known issues about negative feedback.
But Lucio tested a few recent design solid-state low cost pre-amps, and reported that they are nearly miraculous, for their cost, and they are based on op-amps. And as I am, like Lucio, not as much conditioned by schemas or pre-configured ideas as by my ears, I decided to try to design a RIAA stage definitely similar to one of these, just to understand how far one (or better how far I?) could go this way. And as I have no reason to think that the designer has done anything wrong, I just adopted the philosophy reported in the documentation available.
I am talking about the well-known Lehmann Audio Black Cube Phono Pre-amplifier.
Note that the schematics, apart from a few components, have been completely redesigned, without access to original Black Cube schematics (simply copying would not have been of any interest to me as a designer... too easy that way...) but on the other end I think that any designer must be ready to learn from the others, if he wants to achieve good results. You anyway would probably not believe how much of the electronic designer work is just copying and rearranging old circuits...
No kind of competition can anyway be set up between two completely different objects, one studied for industrial production and one that has been designed exclusively for DIY. I did not even take care of making any comparison test between the two. I just wanted to see how far Op-Amp design with reasonable components and costs, following the same components selection logic as tube design, could go, in comparison to tube design itself.
The result, up to now, is a MM prototype, with two separate boards, PSU and audio circuit; both parts are mounted "nearly in air", that is on a PCB without any printed lead, using directly components pins as connection. This allows for very good components stability, while the number of solders is reduced to a minimum.
On the other side the absence of a ground plane for sure does not help in reducing the effect of interfering signals, both internal and external to the board, so that a double sided PCB is probably a better solution (and the only acceptable from an industrial point of view).
Obviously you could merge the two solutions. In this case you must get a single sided board; its single conductive side will be used as ground plane; you must drill all the holes needed for the pins through the board and then mill away the conductive copper layer around the holes in order to avoid improper contacts between pins and ground plane. Up to now I have been too lazy to rearrange everything in this way, so I cannot tell you how great an improvement you can expect... but the prototype seems to work perfectly without any ground plane, so that I do not feel any real need of making any improvement.
There as usual is a mains filter and fuse integrated in the power supply socket for the detachable mains cable. The mains cable is a TNT-Merlino.
I used a slightly oversized PS toroidal transformer (50VA...) with dual 18V RMS output that gives about 20V after rectification.
I first tested a very simple regulated power supply. It worked quite straightforward, but the matching between the two voltages depended exclusively on the matching between the two zeners (it was less than 30mV, but my zeners happened to come from the same reel: another builder could have been not so lucky...). Obviously the remaining ripple was very low, even though the best value I was able to measure with my oscilloscope was around 1mV, but it could have well been irradiated noise (it should have been in the order of microvolts...).
Transistors are BC449/459 (NPN/PNP), low noise audio grade low power transistors.
Then I decided to test another circuit, which is more complex and should grant a better matching between the two halves.
This other set-up works just as fine as the other, but always with resistors taken from the same reel I find a 10mV difference between the two voltages. I had anyway to use a high quality op-amp, OP285 by Analog Devices, as for example with 5534 the offset was very high.
Apart from producing a good number of the best audio integrated circuit available, AD make all their technical documentation available on their site http://www.analog.com. Their devices are relatively easy to find (all the ICs used in this design are available, at least in Italy, also from RS Components).
The transistors can be the same as above or similar ones.
As a matter of measures, I do not find so much difference between the two solutions. As a matter of sound, I would say that the second one is perhaps a little better, a little more soft and clean, perhaps less pinpointing but more musical. But the difference is so small that I doubt of my ears.
There is at least one more solution, that is to connect the input of the inverting block to the output of the non-inverting block, which is by the way the more standard solution, as far as I know. I am going to test this solution too, but I do not expect any big difference. Another way is obviously to use an integrated, low noise voltage regulator, but I am trying to avoid input-output feedback even in the regulator... probably it is just a stupid idea, anyway.
As you see from schematics there is plenty of capacitors; in facts there are nearly 33,000uF. All electrolytic capacitors (normal quality) are bypassed by good quality film capacitors (Philips).
The PSU board is connected to the audio board by a shielded three lead wire; no connector is used, the wire is directly soldered to the boards. The idea is to have a two box device, but the audio stage seems insensitive to such an extent that you could well put everything in one single box, provided it is large enough to leave the two boards far apart (lets say 15cm at least).
I had one major doubt: how much and in which way the PSU unit can change the sound of a device that should have a 80dB PSRR (Power Supply Rejection Ratio, the ratio between a change in power supply and the related change in output voltage), as the op amps I used? As anything seems to have unexpected effects on sound quality, I was pretty sure PSU could change it, but to what extent? Well, initially I was not able to find the suggested ultra fast rectification diodes. Later Norbert Lehmann kindly suggested me again to try that way.
When I finally was able to find and mount them, sound changed dramatically, to a really unbelievable extent. Much more than substituting a film coupling capacitor with a paper and oil one.
The first stage makes use of a very good, low cost, integrated audio pre-amp, SSM2017 by Analog Devices.
SSM2017 is an exceptional pre-amplifier, with ultra low noise and I would say a general very high rejection to any disturb. Probably this is definitely necessary for its normal usage with balanced microphones.
It is in facts designed to work with microphones and their long wires, in the rather electronically polluted environment of a stage or recording session. Common mode rejection ratio is really good, so that it makes very hard for any interfering signal picked up by any reasonable connection wire to be present at the output pin.
Moreover the intrinsic noise of the device is so low that it is normally absolutely hidden by other device's noise (I mean line pre-amp or power amp...).
Second stage is an OP275 dual op-amp in a non-inverting configuration; it is a relatively low cost audio grade amp by AD. Anyway here you can test any other dual op-amp you can put your hands on, requirements are not strict at all. I am currently using an OP285. A fet input op-amp is anyway to be preferred.
The overall gain has been set in order to have 440mV output with 4mV input at 1kHz.
Input stage gain has been set to 40dB, but can be increased for usage with MC pick-ups, if required. Gain is controlled by R2; a table with the relationship between gain and R2 value is present in AD data sheets, so have a look there for changing its value. Anyway you can take input stage gain up to 60dB by taking R2 down to 10 ohms (minimum value in the table).
Output stage gain has been set to 11, but can be easily increased changing R6/R7.
Input impedance is controlled by Rin and Cin: no facility for changing their value is present, because I think that such kind of things can be avoided in DIY design. If you think you need to change the value without using your soldering iron you can use any one of the many arrangements (external load RCA pin, internal switch for load selection,...) that can solve your problem.
I tested the pre-amp with a 4mV output MM pick-up and with an high output (2mV) MC pick-up, but in both systems where I tested it there was no apparent noise produced by the phono stage; they were both tube systems, and one of the two made use of very high sensitivity loudspeakers, that are often able to make evident any amount of noise (in facts, in both cases a slight amount of 50Hz noise was present... but it was independent from the selected source and the noise carpet behind the 50Hz did not changed inserting the phono stage: apparently any noise was due to the following stages....). Hence I don't think there would be any real noise problem increasing the gain ten times.
I have studied the RIAA network adjusting values for an overall error less than +-0.2dB, but I suspect that it is still too high (I repeatedly heard that 0.2% tolerance is not audible, but I am going to suspect it is quite audible...). Anyway this allows a relatively good result with absolutely standard values, which could be a good point in a DIY design.
I decided as usual to implement the network according to the old RIAA standard, not the new RIAA/IEC one. This allows for enhanced very low frequency response, but even rumble is amplified....
I would have liked to test a completely balanced connection to the cartridge, but cartridges normally have their body electrically connected to one ground pin, so the signal just is generated unbalanced, and anyway the SSM2017 product documentation for the pseudo differential configuration gives as a maximum resistance value 10Kohm, while for an MM pick-up I would have needed 23.5K.
One problem I found with the unbalanced configuration is that the first stage output has a DC offset (first stage gain is 1000 at DC...) varying apparently with the cartridge internal resistance, which is someway unpleasant, as the stage could be working not at the minimum distortion point. But this is not sure at all (maybe around 0 volts there could be some crossover distortion that is not present when an offset is applied). There instead is another major issue, which is that the offset is high enough as to require an AC coupling to second stage, which I would have definitely preferred to avoid.
I also had to use one more output coupling capacitor to avoid any offset voltage to be inserted into the grid of the line stage of my tube pre-amp; increasing C13 to 10uF, R6 to 15k, R7 to 150k and decreasing R13 to 13.6k, the offset should get so low, anyway, that this could probably be avoided, in general. Anyway this would require modifying all the RIAA, as it has been designed taking into account the whole circuit, including all coupling capacitors.
For all RIAA and coupling capacitors I used the best capacitors I had at home. They are mostly Selecap (polypropylene), but I used a couple of Jensen paper in oil capacitors as interstage coupling capacitors (I just wanted to try comparing tube with solid state technology with the same components quality level: the eight capacitors I used on signal path and RIAA are rated 630V and all together cost just short of 80$...).
I would normally suggest to use paper and oil capacitors everywhere, if (economically...?) possible, but in this case I tried using paper and oil even as output capacitors, and in the end I came back to film ones, as the sound got a little too much brilliant. Note anyway that all this fine tuning depends really on the character of the other part of the system and your own taste, so you must feel free to select any other capacitor type in order to tailor the sound to your needs.
All resistors are metal film 1%.
Note that power supply filtering is completed on the audio board.
First there is a RC filter on each, positive and negative, supply line (R11+C7A-B and R12+C8A-B). I used a rather low value capacitor at the audio circuit power input, as it should just filter out mainly high frequency noise that could be picked up by the wire connecting PSU and audio boards. The electrolytic capacitors are paralleled with 0.47uF Philips good quality polypropylene capacitors.
Then there is another independent RC filter for each stage and each line, with the IC power supply bypass capacitors, that are set-up as near as possible to the ICs. They are normal quality 4700uF electrolytic capacitors, in parallel with 0.47uF Philips good quality polypropylene capacitors. I tried using lower values for the first ones (220uF, either normal quality or even Elna Cerafine) but there seems to be a residual audio voltage on the power supply pins. This seems to disappear completely if 4700uF capacitors are used.
This all adds up to more than 19000uF on the only audio board.
Even though I was able to test the audio board just by putting it over a grounded insulated metal foil, the board must be completely shielded; best solution would be to use also a PCB with a wide ground plane on the component side. The shield should be connected to the ground near the output connectors (or at least this is the best arrangement in my system...).
Neither input nor output connectors cold side must be directly connected to shield, but their leads should go to the audio board as shown in the schematics. The ground wire coming from the turntable should be connected to the shield on a separate binding post.
The same airiness and depth was present in the sound from my tube pre-amps, but what really is different is that I was used to associate airiness in some extent to a slight thinness of the sound, a little edginess, a lack of low frequencies. Here the sound is as full and solid as anyone could imagine it might be in a solid state system, and at the same time the sound is airy and sweet and liquid. This makes the object work perfectly with both rock and classical music. For sure rock is better handled by this design than by my MW Pre phono stage: I started listening much more rock than before, since I got this working...
And grain, which is one of the typical problems of solid state? Well, I just must say that I have not been able to detect any grain effect that could not be attributed to the record itself. So I am not sure that any grain is really audible: even though I admit that in some situations and with some instruments (strings, typically) something like grain can be heard, it also might be the real string sound, which is often rounded off by not so accurate and detailed reproduction systems.
With regards to dynamics, I can only say that I was caught by surprise, late at night, by a sudden explosion in an old edition of Mahler 1st Symphony (Abbado, DGG): really impressive, for sure speed is not lacking and the impact gets a lot of strength from the very good bass control; the TNT Solidphono has for sure more impact than the TNT MW Pre, even though both have a very good speed.
Instead if we look at microdinamics, the higher transparency of MW Pre01 seems to make it somewhat better. Indeed MW pre is more transparent, but on the other side if you listen carefully you notice that this is mainly due to its rather light bass.
What does this all mean, in the end? Is tube design better than solid state or it is true the opposite? Well, the answer is still difficult. In facts, I like a lot the sound from this new solid state friend, but I like even the sound from MW-Pre. With this last music is always dancing about, it is light, musical, friendly, warm, free. With the TNT Solidphono music is just there, with no frills but full bodied, with a stable, detailed but not excessively pinpointing image, and at the same time with many characteristics one is used to associate to tubes; but I miss a little that magic dancing, that complete freedom.
Much a matter of taste, in the end.
I want to thank Norbert Lehmann of Lehmann Audio for the help and the assistance. For more information about their products and design philosophy please refer to the Lehmann Audio official site.
A regular TNT-Audio reader, Massimo Baso [E-mail: massimobaso (at) yahoo.it], has created images for building TNT Solidphono PCB's and offers these for free to fellow enthusiasts. Contact him for details.
© Copyright 1999 Giorgio Pozzoli - www.tnt-audio.com