Only the regulator has to be as close as possible to the active circuits...unless it uses sense wires.
After a quick listen to the non-oscillating regulator i am pleased to note that the audible hiss has practically disappeared and the Paradise is indeed very quiet.
Sound is quite pleasant and if anything is not quite to my liking it is the spatial aspect - both sound stage width and depth. Not sure if that's the circuit's sonic signature or the pcb groundplane...
That's perfect and it should be so, as HF oscillations are usually audible as increase of hum or hiss or both.
Regarding your LF fluctuations, do you have your boards in shielded boxes?
Not at the moment but will be very easy to have them boxed. Why would the shielding matter? Just blowing on the input transistors generates tens of millivolts at output. Will definitely experiment some more with speeding the servo up.
Just box the input transistors (use a matchbox (for testing) or something like that). If this works, then use a small copper box (so it can dissipate some heat).
I seem to have a bit of a gain mismatch, despite all the components being ruthlessly matched, input transistors better than 5%, current mirrors better than 10% and output from the shunts are all matched within .1v
Any suggestions?
Any suggestions?
That has being discussed before, you can match the gain by adjusting the emitter resistors of the input stage. How much is it ? 1dB ?
'about 1bd?' - it's about 10cm off to the left from my central seated position, so yes. I'll push a reference level through it today and scope the output.
AC voltmeter will be more useful. Just make sure you stay at a reasonably low frequency. Unless you have actual measurements it is impossible to know if the channel imbalance is not due to cartridge, antiskating compensation or something else entirely. As Joachim explained earlier on you can then reduce degeneration from the emitter resistors of the weaker channel by soldering resistors in parallel.
;-) it's not the deck set-up, that is perfectly in balance on all the various stages that pass through my doors. It swaps channels when i swap the outputs from the boards but not when i swap the tonearm leads going into the boards.
I'll create some files on my pc and run them through. I should test the RIAA curves as well and see how close they match.
I'll create some files on my pc and run them through. I should test the RIAA curves as well and see how close they match.
Agree. Frequency should be 1kHz, where sensitivity of the ear / voices etc matter. I would be surprised if you see a lot of gain difference from teh paradise, with this matching.... For example, my "good" cartridge which was a little expensive has 1dB difference between the channels.
In fact, that made me think about a way to adjust the gain balance, e.g. +/- 2dB in steps of 0.25dB or so, what do you guys think?
If it coincides with the cart's imbalance is bad. Better be using a test disc and swap channels for best centering of the whole system. If the RIAA curve is irregularly off across frequencies between channels also, that would mean better to be trimming passives. It takes FFT or a calibrated generator to 0.1dB in audio band. A 1/100 low output impedance resistive divider will be needed to drive the sensitive input and try match to target curve or a precise anti-RIAA and try match to flat.
Yes, that's fine, but:
How to do it in a very "practical way" (with al the emitter resistors soldered in the double side print pcb)."
How to measured the "inbalance" between each Tr BC337 & each Tr BC327 ... and adjust in fine & practical way the emiter the resistors).
Thanks for (practical & expierences) ideas (method)..
I did ask about this a long long while ago a single point where to make small changes would be really useful.
About the emitter resistors there are 8 of those so....
I was thinking about a trimmer resistor to change the value of the resistors on one of the branches of the current sources I have not tried this so far and I am thinking that If I change one side, for example the positive side it will alter/shift the offset so maybe this will require a 2 point adjustment.
So I was going to try 2 X 10K trimmer soldered across R3 and R7 or maybe better R4 / R8 as placing a trimmer in parallel reduces the overall value.
As you can gather I am fishing in the dark and as I am learning a good explanation would be most welcome
Or maybe R12 and R13 with a 100 k trimmer but this would still be a 2-point.
A parallel line of thought follow this, would shifting the level here help reduce the offset and therefore reduce the work load to the servo?
Simon what are your voltages at the test points?
Could be that one of the resistors is slightly out?
Another reason I am looking elsewhere apart from emitter resistors is that if there are component tolerances affecting the performance those should average out as I said there are 4 emitter resistors on each branch.
The idea of Ac voltmeter against oscilloscope is totaly @@@@@ Voltmeters have limited band width and low imput impedance compared to a scope besides even my crapy scope has 2 channels and sum and subtract functions. But this is just more dross....
Cart Imballance with Simon set up I do not think so I have seen it😀
Last edited:
Certanly this is a more elegant solution (swapping boards)
I was thinking about a passive RIAA (EG hagherman) connected to both channels at the same time if there are errors on the Passive RIAA those will be common and not efect the mesurament of the difference on the gain of the 2 channels.
By the way how is Salas Paradise camming along?
This is probably the build I am looking out for before I start with my R3
This is probably the build I am looking out for before I start with my R3
left channel
VCASCL 12.38
VCASCH 14.37
right channel
VCASCL 14.35
VCASCH 14.40
The transistors were a good match, +/- 10hfe.
I thought I had matched the resistors between branches, maybe I didn't.
Also RIAA EXT is 2.3vDC on the left and 3.0vDC on the right.... ?
VCASCL 12.38
VCASCH 14.37
right channel
VCASCL 14.35
VCASCH 14.40
The transistors were a good match, +/- 10hfe.
I thought I had matched the resistors between branches, maybe I didn't.
Also RIAA EXT is 2.3vDC on the left and 3.0vDC on the right.... ?
Last edited by a moderator:
VCASLC 12.38 then is possible culprit
And Sallas if you are stuck for transistors I have started BCXXX40 tread
in the swap forums
This is intended to be strictly a non profit orientated tread
Post what you have and what you need in there in regardsto the transistors Hfe
hopefully it will pick up...
http://www.diyaudio.com/forums/swap-meet/226166-paradise-bcxxx-40-a.html
And Sallas if you are stuck for transistors I have started BCXXX40 tread
in the swap forums
This is intended to be strictly a non profit orientated tread
Post what you have and what you need in there in regardsto the transistors Hfe
hopefully it will pick up...
http://www.diyaudio.com/forums/swap-meet/226166-paradise-bcxxx-40-a.html
