Cripes! I didn't think my question would end-up producing such a long thread... the discussion has been interesting, and thanks for the offer of help PowerRex.
I am by any stretch of the imagination an electronics noob, but I definitely can wield a soldering iron, so not quite scared away... The disucssion has set me thinking - prehaps reducing the gain of the preamp (which is a heavily modified Naim 62) would be better, as sugegsted by one poster.
I suppose though that that would mean that the phono stage (a relaid version of the Naim circuit which has 4x separate regulators powering its various stages) would need extra gain. Having said that, the cartridge (Troika) is very low output so a more modern one might fix that issue too...
I haven't looked inside the DAC for ages, but will do so as soon as I can - I know the op amps are different from the standard Philips ones...
J
I am by any stretch of the imagination an electronics noob, but I definitely can wield a soldering iron, so not quite scared away... The disucssion has set me thinking - prehaps reducing the gain of the preamp (which is a heavily modified Naim 62) would be better, as sugegsted by one poster.
I suppose though that that would mean that the phono stage (a relaid version of the Naim circuit which has 4x separate regulators powering its various stages) would need extra gain. Having said that, the cartridge (Troika) is very low output so a more modern one might fix that issue too...
I haven't looked inside the DAC for ages, but will do so as soon as I can - I know the op amps are different from the standard Philips ones...
J
Let me get this clear: your CD has too much output for your line stage, so you reduce the gain of your line stage, then you need to boost the gain of your phono preamp and/or change your cartridge. It seems a lot of effort for a problem which can be solved with four resistors (two per channel)! (You already have resistors in your audio path - its called a volume control).
If you are happy to mod your equipment, why not simply bypass the line stage for CD input? That way you reduce noise and distortion too.
If you are happy to mod your equipment, why not simply bypass the line stage for CD input? That way you reduce noise and distortion too.
Hi professor,
James1 is absolutely right to reduce his preamp's gain : if you read carefully his original post, for a given position of the preamp's potentiometer (even at the lower positions) the CD is too loud and the LP is too low. So, if you just decrease the CD's output voltage to adjust it equally to the phono source, the user must always put the volume control at high positions to have some reasonable sound level : by the way, the dividing resistive network is not a good solution for sound quality as described before in several posts by different members. Again, if you read James1's initial post carefully, this solution that every single young electronics amateur would suggest is indeed very cheap and simple to apply but it degrades signal integrity : yes, the preamp's attenuator acts like your preconized voltage divider but we all know that this is a very strategic part in the signal path simply because potentiometers alter sound quality !
If it's mandatory to use 1 at least, let's avoid using a second one...
As digital sources generally give 2-3 VRms (except some few high output level DACs like the DCS Elgar...) for a 0 dB recording, it's not sure at all that James1's DAC has a too high output voltage : you'll of course claim, as always with bold statements, that it's largely enough to saturate any amplifier's input but, in practice, this is not the case as most quality recordings never reach 0 dB, otherwise the DAC chip's output would saturate ! So, in order to keep some headroom margin for high-dynamics audio signal levels, the average output voltage is around the hundreds of millivolt range most of the time, and it reaches the volt range for some short periods. Of course, you can find junk recordings where the DAC output is often saturated but they are rarely listened to by audiophiles because of their poor sonic performance.
It seems that you are here to force James1 to chose absolutely your solution, which as always you consider as the only best one, but keep in mind that he has to cope with his own system's issue to find a satisfactory compromise !
I'll also reply to your last quote.
Have fun !
This is really great so I thought it ought to be repeated. Perhaps it should be engraved on a bronze plaque and hung on the forum somewhere.
5xRC is an approximation, as you seem to forget : charging and discharging formulas [Vs(1-e(-t/RC) and Vs(e(-t/RC)] need a calculator near your hands for precise design simulations so we generally admit that a capacitor is charged approximately after a 5xRC delay : with a 10 V supply voltage, the cap is considered as charged with 9.93262 V between its electrods ! Do you search for the complete Pi value before calculating a circle's area ? No, it's impossible as Pi has an infinite number of digits so you approximate it to 3.14 ! Oh sorry, 3.141592654...
Yes, every filter will alter signal integrity and, in Hi-Fi, we always talk about dynamic signal where frequency, phase and amplitude ALWAYS varie in an unknown manner : that's why theoretical masterminds like you never understand the influence of time in practice.
Concerning jitter, what I meant was that every filtering network in the analog domain will cause similar time mismatch phenomenons to the dynamic signal as does jitter in the digital domain which can be compared to a timing error : jitter is indeed random, otherwise I'd rather compare it to a delay line as you seem to present the RC network ! But do you know that even delay lines's value is given for a narrow frequency bandwidth ? So, when you pretend that the delay induced by RC is fixed and linear, it's completely faulse : it's related to the considered frequency. The phase response is not a straight line !
If you talk again about the 80 Khz's cut-off which is too high compared to the maximum 20 Khz of a CD, OK if we're dealing with a static sine signal : again, you forget that in the real world, two or more spikes could occur in a very short time, for example during a heavy and fast hit on Cymbals. In this case, what is the equivalent frequency to deal with ? Always no phase problems with your RC network ?
Concerning the damping factor, I was just comparing low source impedance feeding high load impedance cases with the damping factor of an amplifier's (low output Z) driving a speaker's (higher Z) which value, at a given frequency, is Zload divided by Zsource : higher damping factor leads to a better cone control. By comparision, and comparison only, I meant that in both cases, source impedances are low vs load impedances.
Concerning my studies, they seem to give very satisfactory sonic results in the audio field since near 20 years : but conventional designers always disagree theoretically and are very positively surprised when they listen to the practical results ! So somebody is certainly wrong...
why don't you use a 5k or 10k pot between the CD and preamp to find the proper level. (R1) Then measure the wiper to GND resistance and input(CD) to wiper. (R2) Try to keep the total series resistance (R1+R2) below 10k and above 5K to balance load and bandwidth concerns
Then buy 1% metal film resistors in the most correct ratio and solder them into an adapter connector.
Then buy 1% metal film resistors in the most correct ratio and solder them into an adapter connector.
No, I'm not a professor (just a PhD, and that not in audio but radio - but so what? Truth is truth, whoever says it. What sort of person uses 'professor' as a form of insult?).
I fail to see how 5xRC is a good approximation for infinityxRC (time for ultimate settling) or 1xRC (the actual time delay for most signals). All this loose talk about signal smearing from a source which has already been through two near-brick-wall filters would be funny if people did not take it so very seriously! A single-pole low-pass filter acts like a pure time delay equal to RC for all frequencies significantly below its turnover frequency i.e. where there is little attenuation. 20kHz is sufficiently below 80kHz for this to apply. A pure time delay cannot damage a signal. You would be unlikely to notice even if this was applied to just one channel. If you don't like my suggestion you could follow SY instead; the principle is the same.
As I said much earlier, if you don't trust resistors then you must go pure acoustic for your music. This discussion is now going round in circles, so I have had enough.
I fail to see how 5xRC is a good approximation for infinityxRC (time for ultimate settling) or 1xRC (the actual time delay for most signals). All this loose talk about signal smearing from a source which has already been through two near-brick-wall filters would be funny if people did not take it so very seriously! A single-pole low-pass filter acts like a pure time delay equal to RC for all frequencies significantly below its turnover frequency i.e. where there is little attenuation. 20kHz is sufficiently below 80kHz for this to apply. A pure time delay cannot damage a signal. You would be unlikely to notice even if this was applied to just one channel. If you don't like my suggestion you could follow SY instead; the principle is the same.
As I said much earlier, if you don't trust resistors then you must go pure acoustic for your music. This discussion is now going round in circles, so I have had enough.
Now James1 just has to check each suggested solution :
- the cheapest one would be to use the resistive dividing network : not before but after the interconnect cable (just at the preamp's CD input) to avoid coaxial cables' capacitive influence . In this case, the phono and CD levels being matched, the global level evolution still needs a far bigger raising of the volume control, as initially for the phono section.
- the most coherent one would be to increase Rg in the preamp's negative feedback network (Rg goes from (-) input to ground) to lower its gain, then to rise up the phono section gain and leave the CD's output stage as it was initially. In this case, there will be enough rotation margin for low and high listening levels for both CD and phono inputs. Lowering slightly the preamp's gain won't affect its stability, the voltage noise will decrease and the bandwidth will also rise slightly.
Both solutions are very easy to test and James1 can then tell us more about the one with the best sonic results, in practice. No need to be a PhD for it...
- the cheapest one would be to use the resistive dividing network : not before but after the interconnect cable (just at the preamp's CD input) to avoid coaxial cables' capacitive influence . In this case, the phono and CD levels being matched, the global level evolution still needs a far bigger raising of the volume control, as initially for the phono section.
- the most coherent one would be to increase Rg in the preamp's negative feedback network (Rg goes from (-) input to ground) to lower its gain, then to rise up the phono section gain and leave the CD's output stage as it was initially. In this case, there will be enough rotation margin for low and high listening levels for both CD and phono inputs. Lowering slightly the preamp's gain won't affect its stability, the voltage noise will decrease and the bandwidth will also rise slightly.
Both solutions are very easy to test and James1 can then tell us more about the one with the best sonic results, in practice. No need to be a PhD for it...
PoweRex - did you join up this month just to give us all wild and impractical solutions to easy problems?
There are many members on this forum with solid engineering chops. Blasting in here with difficult solutions to easy problems is not very polite. Especially when good, cheap, practical advice has been offered. Perhaps you should read the forum a bit and get a better feel for it. Enthusiasm is welcome. Running rough shod is not.
James. Try the simple resistive divider. It is likely all you will ever need.
There are many members on this forum with solid engineering chops. Blasting in here with difficult solutions to easy problems is not very polite. Especially when good, cheap, practical advice has been offered. Perhaps you should read the forum a bit and get a better feel for it. Enthusiasm is welcome. Running rough shod is not.
James. Try the simple resistive divider. It is likely all you will ever need.
Dear Panomaniac,
If you read my last post completely, you'll notice that I suggested to James1 to test both solutions and to make his own choice !
But it seems clearly that you force him to chose yours...
Both solutions are easy to test so let James1 chose, that's it.
Have fun !
No, I didn't !
This is a part of James1's first post :
"Is there a way to reduce the gain of the audio output of this circuit *without* simply using an attenuator, which might reduce the output but also kill musicality."
So, I just suggested him an alternative solution to cope with his level issue, that's all. Now, he'll chose by himself.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.