The preamp is now functionally complete, just waiting for the final coats of paint on the endcheeks and for the front panel to be anodised. This has been a very lengthy project, with more than five years passing between me putting down my design notes and the final installation of the preamp into my system. In the meantime, of course, I was occupied with other projects besides the preamp - I've also built my two stereo high-speed power amps and rebuilt my active crossover.
Most of the contruction was relatively straightforward, with most of the time being taken up with trying to sort out two major bugs in the volume control.
After many months of trial and error, I made a couple of changes from my initial plans.
I ultimately failed in my attempts to feed a three-wire signal to the DS1802 from a TTL circuit. Although I'm sure that this is theoretically possible (and vital if I upgrade to a different chip such as the Crystal CS3310), I think I'll need an oscilloscope to debug the logic circuitry. I resorted in the end to feeding the control signals from the front panel directly to the "push-button" inputs of the volume control chip. I also had to add an extra 74LS32 chip to ensure that the mute indicator was reset whenever I changed the balance or volume levels, since this is what the volume control chip does.
I also gave up trying to make my homebrew optical rotation detector on the volume control shaft work. I found that the IR opto detectors I used were inordinately sensitive to the choice of paper and ink on the shaft-mounted drum; I eventually gave up and bought a Bourns rotary encoder, which has the additional advantage of providing a good-quality bearing for the far end of the shaft (my initial arrangement simply felt loose). The encoder I used had 64 pulses per revolution, which matches well with the 64-step volume control.
I had to completely rethink how to do this with the pushbutton inputs of the DS1802, since using each of the two chips in balanced mode meant that I had to feed separate volume control signals to each stereo channel (if I had used the chips in single-ended stereo mode I could have used the balance controls on the DS1802). I managed to implement balance control by adjusting the level of the two channels in opposite directions when the balance was changed.
My initial guess at increments of +-1dB for each channel turned out to be justified: each step of the balance control gives a noticeable change in position of a central image, and the extreme settings of 6dB to the left or right result in the image almost collapsing into one speaker or the other.
This is a vital consideration when using the DS1802 volume control chip, since the low supply voltage on this (5V maximum) imposes a maximum signal level of 1.76V RMS before it runs unpleasantly into clipping. To start with I set up the preamp with all the inputs except the CD input going straight into the input of the DS1802, and with a 10dB gain in the output stage. The redbook specfication for CD gives a maximum output of 2V RMS, which is about 4dB above the standard 0dB line level of 0.775V, and 1dB above the clipping level of the DS1802. I initially intended to use a 10dB cut on the CD input to bring it down to unity gain overall, but I settled for 6dB in the end, which brought it subjectively to the same level as the FM tuner input.
This setting worked fine for all sources except for black disc, which produced loud cracking noises every time the stylus found a speck of dirt. With the passive preamp I was using before, the decent overload margins on the phono stage and the high slew rate of the power amp meant that the system was very well-equipped to deal with this kind of transient. However, even though the music level from the phono stage shouldn't cause any problems, the occasional high-speed transient in the surface noise cause the DS1802 to clip; although these don't necessarily happen that often, their subjective effect is pretty unpleasant. Adding 10dB of attenuation to the phono input fixed this problem for all but the very grungiest records..
I'm rebuilding the remote so that it allows access to all the functions on the front panel except the record select, and also will come in a much more robust and domestically acceptable form.
I toyed with the idea of using a different chipset for the remote, partly because the MV501/MV600 are obsolete, and partly because they need a 9V supply and the PP3 batter is bulky and heavy. In the end I stayed with the MV501/MV600 since I couldn't find a readily available substitute that did the same job, and I know it ain't broke!
I was relieved to find that hum and noise levels are subjectively very low, and that the zero-crossing on the volume control works completely imperceptively. In fact in all respects the preamp behaves itself impeccably.
To my great disappointment, unfortunately, even after running in for a week it sounds a little uninspiring in comparison with my old Croft Super Micro A, which brings rather more energy, life and presence to music. Puzzlingly, it sounds quite bearable when I'm listening to CD or the radio, but rather "hi-fi" and musically uninteresting when I put on an LP. It seems ironic that, despite the claims for the technical superiority of digital media over the near-obsolete black disc in terms of signal-to-noise ratio and frequency response, the latter medium seems to be rather more demanding of the frequency and dynamic range of the equipment in the rest of the signal chain. Whether or not any true musical signal is recorded on an LP above 20kHz, there is usually plenty of high-frequency energy coming out of the cartridge pins, particularly with a modern cartridge like my Lyra, with its tip resonance of 50kHz or so. These fast transients can have a high peak level, so the phono stage and other circuitry downstream needs to have both a highish slew rate and a good dynamic overload margin to ensure it doesn't get confused by surface noise.
I suspect that the volume control chip is the main culprit, with its limited dynamic range and significant (if not huge) distortion. It isn't intended for high-resolution audio use at all, and I suspect is aimed at applications like TV sets. At the moment I'm investigating the possibility of using motorised potentiometers to get the same functionality as the DS1802, but with a much larger dynamic overhead.
Alex Megann, November 2002
Update: January 2003
I've decided to implement my original plan from more than twelve years ago and go for a relay-switched attenuator with 2dB steps. This is pricey and complex (especially with the balanced operation doubling the number of channels up to four), and will take several months to plan and collect the parts for.
In the meantime I have replaced the DS1802s with a couple of dual 10K Sfernice P11 pots ganged together, with 2K resistors between each wiper and ground to make a (slightly) better approximation to a logarithmic law. With these in place the preamp sounds immediately better with all sources than with the digital pots - whether it is my imagination or not, they sound much more open and dynamic, and vinyl in particular sounds very satisfying. The main practical drawbacks are twofold: firstly, I lose the nice logarithmic attenuation, which means that control at low volumes is hit-and-miss. Secondly, the Sfernices are widely described as good-sounding but with poor channel matching, and I can confirm this. Even after matching the pots into pairs with a multimeter, as I raise the volume from zero the balance swings from left to right and then back to the centre again.
I had to reduce the preamp output gain from 10dB to unity to get better control at low levels, but this in turn introduced a small DC offset into the system which produced clicks and pops when I changed sources or switched into standby (presumably from the ground, rather than signal, switching). I'm looking forward to getting my relay attenuator into action...
Update: April 2005
"Several months"? Hmm - perhaps overambitious... A house move and various other events got in the way, so I've only just finished rebuilding the preamp board onto proper PCBs. The opamps are now soldered on directly instead of being fitted in sockets, and I have used precision resistors everywhere. The Sfernice volume pots remain, as I'm still soldering stuff onto the attenuator boards.
While the preamp was on the workbench I substitued my old Croft Super Micro A, which turned out to be an interesting comparison. I was initially very impressed by the Croft on first audition and paid nearly £700 for it - it worked very pleasingly with the small Royd and Shan speakers I used it with, through one or two of my Croft Series 5 valve power amps. Indeed, when I first had a Croft preamp (a Super Micro Mk. 2) the combination with a Quad 306 and Royd Eden speakers it was a high point of my audio history, even if big orchestral music was well beyond the abilities of the little speakers. The Croft had a fuller bass than my OPA2604-based preamp and sounded invitingly warm, but I was surprised by how much detail and texture that was naturally present with my homebrew preamp was missing with the valve preamp. There was a vagueness that was irritating on longer listening sessions, even though the Croft was generally very easy on the ear, with natural dynamics and a nice funky feel to rhythms. I do recall years ago that a passive preamp (OK, a 10K potentiometer) sounded similarly clearer than the Croft, so it's not simply a matter of ageing valves.
I owe, as usual, many thanks to Neil Jenkinson for letting me use his PCB etching facilities.
Update: September 2005
I have just reinstalled the preamp into my system after a spell of about three months where my old passive preamp was standing in for it while I soldered in the relay-switched attenuator boards. Comparison with the passive was really interesting. When I took the active preamp out and put the passive in, I felt that the sound lacked body, dynamics and tonal colour. However, as time went on I realised that with the passive the system was straining, which was a surprise to me. With relatively recent, highly multi-tracked digitally recorded LPs such as Pink Floyd's "A Momentary Lapse of Reason" the higher frequencies easily became congested, making the system tiring to listen to. Big orchestral works also sounded bleached and unexciting, even at higher volumes.
When I put the active preamp back into the system everything seemed to come to life again, and I found the same Pink Floyd album much more satisfying and easier to listen through - in fact, I got the the end of the first side and started listing to side two before I ran out of time, where a week earlier I had had no desire even to finish side one.
Why a passive preamp should sound distorted in this way and rob the music of its joie de vivre I have no idea. A switched attenuator has negligible intrinsic distortion, and I had made a point of cleaning all the switch contacts before putting the passive into my system. Loading my CD player and phono stage with 10K shouldn't be a problem of any kind. My only thought is that some RF may be getting into the system while the impedances are high and interfering with the power amp, though this ought to be affected less than most solid-state power amps - partly because of the high slew rate, and partly because of the input filter which should weed out most things higher than a megahertz.
Anyway, I am very pleased with my relay attenuator. In terms of practicality, it is definitely a huge improvement over the digital chip. Although I couldn't convince myself that the sound was vastly better than with the Sfernice pots, the practical aspects are very much improved. To my relief, there are no clicks as I change the level, and the volume control is much more practical than the old Sfernice pots - the balance at low levels is perfect, and I find that I am using a much wider range of volume settings (though I do think that the gain of my phono stage is about 6dB too high for my Lyra cartridge).
Back to DIY page.