mega lord
Centre tapped
Should I remove the magnets from my speakers?
Pete
Yeah, why not
I was thinking having an EI transformer right next to the tray magnet (if i put it back) might be impeding the operation of the magnet. Can't be having that
Naim CD3.5 mods (not on Acoustica)
- Thread starter mega lord
- Start date
Yeah, why not
I was thinking having an EI transformer right next to the tray magnet (if i put it back) might be impeding the operation of the magnet. Can't be having that
mega lord
Centre tapped
It rained here on and off all day so i decided to fanny about with the 3.5 a bit more and phone LesW about something else.
In post 123 there are pics of the two blue pcb EI transformers i have powering the DAC and flea powered clock. These were salvaged from some old gas boiler pcb's. They have 3 secondary windings, 0-10V @ 2VA and two 0-9V @ 1.5VA each.
When i initially decided to use these i thought i had to put the two 9V windings in series to have enough voltage for the flea powering the tentlabs clock and the flea on pin28 of the dac. Theflea's require 18V minimum in put voltage . Putting the two 9V windings did give me over 30 Volts dc.
I finally plucked up the courage to try and use the 0-10V winding to power the flea on dac pin 28. I thought that as the regulation was so poor on these tiny transformers that maybe i would have enough voltage after all. This is what i have done and left dac pins 1 and 10 powered from the original 9V windings in series. This would use the transformer's full VA rating (5VA) and give the flea and dac pin 28 its very own transformer winding and 0V reference.
I was a little worried at first that the ten volt winding would not give the magic 18V required for the flea. But as it happens the lowet voltage that i got today was 19V ! Hooray. Usually the mains is over 240V here but in the middle of the day it got as low as 234V. That is when i measured the lowest dc voltage to the flea of 19V. So it looks like it is a goer.
I used an old bridge rectifier from a nap140, two 1R power resistors between rectifier and cap positive and negative terminals (thanks MC) and a 10,000uf 50V cap for the new supply. Well everything works, no sparks, no magic smoke and the dac did not lock out due to psu start up. Happy days. I used to have a little lm317 pre reg on the dac as the voltage was a little high from the two 9V windings in series. I left that on the rail supplying dac pins 1 and 10 (goes through flea style gyrators and acoustica style led 317 regs before getting to pins 1 and 10) but wired the new supply straight to the flea powering dac pin 28 as i could not
afford the pre regs dropout.
Well the sound has moved on a step further and is rather spiffy now. Who would have thought that totally separating dac pin 28 from pins 1 and 10 would bring such big rewards. I was gob smacked when i isolated the dac and gave it it's own transformer. This was not as rewarding as that but well worth the effort non the less. More detail, textures and round earth stuff without losing what i liked about the 3.5 when i bought it.
I think Naim (with the cdx etc and xps) give the dac its own winding on a common trafo. I think they missed a trick by not giving it its own transformer.
Next mod for my 3.5 will be to separate the two 9V windings and give pins 1 and 10 their own winding, psu and 0V reference. The two 0-9V windings will give plenty of voltage overhead for the gyrators and acoustica led 317's on pins 1 and 10. Also with less voltage to burn the dac lm317's should run cooler and prolong life.
Some of the 317's on the back panel are dropping the voltage from 26 to 5V. That can't be good for them and must be part of the reason they get so hot (along with the current draw from their prospective loads). I might have to try pre regin' them to give them an easier time. As the highest voltage i need from any of the internal regs now ( anologue stage, dac pin28 and inverter 317's have all been removed and internal alwsr on opamps is from an external supply) is 9V, then i can afford to drop the supply voltage to say 15V without any mishaps. Maybe a lower voltage transformer would be a better idea rather than making heat by putting a pre reg in.
Big grin on my face at the moment.
Stu
In post 123 there are pics of the two blue pcb EI transformers i have powering the DAC and flea powered clock. These were salvaged from some old gas boiler pcb's. They have 3 secondary windings, 0-10V @ 2VA and two 0-9V @ 1.5VA each.
When i initially decided to use these i thought i had to put the two 9V windings in series to have enough voltage for the flea powering the tentlabs clock and the flea on pin28 of the dac. Theflea's require 18V minimum in put voltage . Putting the two 9V windings did give me over 30 Volts dc.
I finally plucked up the courage to try and use the 0-10V winding to power the flea on dac pin 28. I thought that as the regulation was so poor on these tiny transformers that maybe i would have enough voltage after all. This is what i have done and left dac pins 1 and 10 powered from the original 9V windings in series. This would use the transformer's full VA rating (5VA) and give the flea and dac pin 28 its very own transformer winding and 0V reference.
I was a little worried at first that the ten volt winding would not give the magic 18V required for the flea. But as it happens the lowet voltage that i got today was 19V ! Hooray. Usually the mains is over 240V here but in the middle of the day it got as low as 234V. That is when i measured the lowest dc voltage to the flea of 19V. So it looks like it is a goer.
I used an old bridge rectifier from a nap140, two 1R power resistors between rectifier and cap positive and negative terminals (thanks MC) and a 10,000uf 50V cap for the new supply. Well everything works, no sparks, no magic smoke and the dac did not lock out due to psu start up. Happy days. I used to have a little lm317 pre reg on the dac as the voltage was a little high from the two 9V windings in series. I left that on the rail supplying dac pins 1 and 10 (goes through flea style gyrators and acoustica style led 317 regs before getting to pins 1 and 10) but wired the new supply straight to the flea powering dac pin 28 as i could not
afford the pre regs dropout.
Well the sound has moved on a step further and is rather spiffy now. Who would have thought that totally separating dac pin 28 from pins 1 and 10 would bring such big rewards. I was gob smacked when i isolated the dac and gave it it's own transformer. This was not as rewarding as that but well worth the effort non the less. More detail, textures and round earth stuff without losing what i liked about the 3.5 when i bought it.
I think Naim (with the cdx etc and xps) give the dac its own winding on a common trafo. I think they missed a trick by not giving it its own transformer.
Next mod for my 3.5 will be to separate the two 9V windings and give pins 1 and 10 their own winding, psu and 0V reference. The two 0-9V windings will give plenty of voltage overhead for the gyrators and acoustica led 317's on pins 1 and 10. Also with less voltage to burn the dac lm317's should run cooler and prolong life.
Some of the 317's on the back panel are dropping the voltage from 26 to 5V. That can't be good for them and must be part of the reason they get so hot (along with the current draw from their prospective loads). I might have to try pre regin' them to give them an easier time. As the highest voltage i need from any of the internal regs now ( anologue stage, dac pin28 and inverter 317's have all been removed and internal alwsr on opamps is from an external supply) is 9V, then i can afford to drop the supply voltage to say 15V without any mishaps. Maybe a lower voltage transformer would be a better idea rather than making heat by putting a pre reg in.
Big grin on my face at the moment.
Stu
martin clark
pinko bodger
The reason to keep all the Vdd pins on a dac on the same supply, is that many bitsream dacs will 'lock-up' if the supply pins do not start-up at comparable rates that keep the pins within a certain voltage of each other
About 15yrs ago the then -popular CS4390 was a bugger for this, if the 5v pins didn't stay within about 0.3v during start-up it would lock up and make no sound until you got it right. But it's very much something that depends on the exact internal design arrangement. I'll need to see if such a requirement is given in the TDA1305 datasheet, but its good practice anyway.
Since you are in control of the design you have and if the fleas in use here have similar filter time constants it shouldn't be a problem at all. But I can understand for several commercial reasons why perhaps Naim didn't bother (such as - cost of a separate supply vs player price point, and other little design things already left hanging like acoustica mod #7 which swamp this...)
PS don't worry about 3pin regs getting hot so long as it's not finger-burningly so; transistor hfe improves with temperature, so jellybean regs tend to perform a little better when warm than not (more effective feedback from more internal forward gain - providing the temp rise doesn't ruin the internal voltage reference performance, or dramatically increase noise.) Weird but true.
About 15yrs ago the then -popular CS4390 was a bugger for this, if the 5v pins didn't stay within about 0.3v during start-up it would lock up and make no sound until you got it right. But it's very much something that depends on the exact internal design arrangement. I'll need to see if such a requirement is given in the TDA1305 datasheet, but its good practice anyway.
Since you are in control of the design you have and if the fleas in use here have similar filter time constants it shouldn't be a problem at all. But I can understand for several commercial reasons why perhaps Naim didn't bother (such as - cost of a separate supply vs player price point, and other little design things already left hanging like acoustica mod #7 which swamp this...)
PS don't worry about 3pin regs getting hot so long as it's not finger-burningly so; transistor hfe improves with temperature, so jellybean regs tend to perform a little better when warm than not (more effective feedback from more internal forward gain - providing the temp rise doesn't ruin the internal voltage reference performance, or dramatically increase noise.) Weird but true.
mega lord
Centre tapped
Thanks for popping by Martin (again).
Top tip regarding regulator heat and performance. I will leave well alone then.
I remember you mentioning TDA1305 start up issues a while back which is why i have been wary of messing with it. it would seem i have got away with it so far but if it does ever lock up at least i know it can be rectified by getting it right (chip wont blow up).
Stu
Top tip regarding regulator heat and performance. I will leave well alone then.
I remember you mentioning TDA1305 start up issues a while back which is why i have been wary of messing with it. it would seem i have got away with it so far but if it does ever lock up at least i know it can be rectified by getting it right (chip wont blow up).
Stu
mega lord
Centre tapped
I carried out some more surgery today.
I split the two 9V windings in series in to two separate windings. I made up another rectifier and 4700uf cap this time as it is all i had left. Rectifier is a small round one that used to be in the preamp psu of a nap140. Again i put a 1R resistor between rectifier and cap positive and negative terminals.
I removed the little lm317 prereg i was using on dac pins 1 and 10 and wired straight from the psu to the flea style gyrators which proceed the acoustica led style 317's on pins 1 and 10. As the voltage supplied to the final regs has now almost halved i did not need the voltage dropping prereg and its current draw has now been taken away. The pre reg had an 0V return to the ground plane near the dac. This has been removed and each of the three transformer windings has its own 0V return which eventually goes to the decoupling caps right next to the dac.
Everything started up as it should. The new 4700uf cap has not been used for years and i witnessed the sound improving as more discs were played. The sound is super detailed now with wonderful textures. I would say the difference in sound from giving pin1 and pin 10 their own winding and psu to be very small but still evident. Not as big as giving pin 28 and the flea it's own psu and winding which in turn was not as big an improvement as giving the dac its own transformer (all 3 supply pins sharing 1 psu).
All hooked up the 9V windings were giving 18V un loaded which dropped to 15V when loaded.
I split the two 9V windings in series in to two separate windings. I made up another rectifier and 4700uf cap this time as it is all i had left. Rectifier is a small round one that used to be in the preamp psu of a nap140. Again i put a 1R resistor between rectifier and cap positive and negative terminals.
I removed the little lm317 prereg i was using on dac pins 1 and 10 and wired straight from the psu to the flea style gyrators which proceed the acoustica led style 317's on pins 1 and 10. As the voltage supplied to the final regs has now almost halved i did not need the voltage dropping prereg and its current draw has now been taken away. The pre reg had an 0V return to the ground plane near the dac. This has been removed and each of the three transformer windings has its own 0V return which eventually goes to the decoupling caps right next to the dac.
Everything started up as it should. The new 4700uf cap has not been used for years and i witnessed the sound improving as more discs were played. The sound is super detailed now with wonderful textures. I would say the difference in sound from giving pin1 and pin 10 their own winding and psu to be very small but still evident. Not as big as giving pin 28 and the flea it's own psu and winding which in turn was not as big an improvement as giving the dac its own transformer (all 3 supply pins sharing 1 psu).
All hooked up the 9V windings were giving 18V un loaded which dropped to 15V when loaded.
mega lord
Centre tapped
Next I think i am going to embark on mods to the ALWSR as described by John Luckins. And then use remote sensing as when i was first installed them in 2007/8 i was a newbie and a bit scared. Got the courage now.
Other things i have been thinking about
1) Has anyone ever powered each anologue stage opamp with a flea ?
Can the flea do 20V output at the load an opa627 or 604 pulls ? Too much heat ? I assume a bigger reg is needed e.g 7824 and other component swaps - output voltage setting resistor divider ?
Would it bring any gains over 2 ALWSR ? Cost is not an object here.
2) Naim use split rail supplies in their higher end players. Can the opamps in the 3.5 be converted to split rail supply ? Are there any gains with this approach and Is it worth the hassle ? Could it be done just for the 6 opamps any nothing else ?
3) Naim use various low value (under 5 ohms) resistors between regs and the three dac supply pins and the four decodersupply pins. Is this just for damping the output of the 317 regs or some other reason. Not all pins have the same value used so they must have picked them for some reason. I removed the one on dac pin 28 when i installed the flea.
Too many questions !
Stu
Other things i have been thinking about
1) Has anyone ever powered each anologue stage opamp with a flea ?
Can the flea do 20V output at the load an opa627 or 604 pulls ? Too much heat ? I assume a bigger reg is needed e.g 7824 and other component swaps - output voltage setting resistor divider ?
Would it bring any gains over 2 ALWSR ? Cost is not an object here.
2) Naim use split rail supplies in their higher end players. Can the opamps in the 3.5 be converted to split rail supply ? Are there any gains with this approach and Is it worth the hassle ? Could it be done just for the 6 opamps any nothing else ?
3) Naim use various low value (under 5 ohms) resistors between regs and the three dac supply pins and the four decodersupply pins. Is this just for damping the output of the 317 regs or some other reason. Not all pins have the same value used so they must have picked them for some reason. I removed the one on dac pin 28 when i installed the flea.
Too many questions !
Stu
mega lord
Centre tapped
Did some more digging around the 3.5 circuit today.
It would appear now that the LM317 in the swing out tray gets its power from one of the 9V regs on the back panel - the third one along from the on/off switch. So the reg in the swing out tray only has to drop 4V not 21V as i previously thought and is effectively pre-regulated. I wonder if Naim did this just so the 317 in the tray does not get too hot ?
Also i mis-traced the output from the second lm317 on the back panel. It also goes to the swing out tray via an RC filter and connects to the sensor (tells the player when the door has been closed) plus one or two other things i have not worked out yet.
The circuitry below the output relay still puzzles me.
Also the relay has an 0V connection right back to near the dac. Still dont understand why.
Stu
It would appear now that the LM317 in the swing out tray gets its power from one of the 9V regs on the back panel - the third one along from the on/off switch. So the reg in the swing out tray only has to drop 4V not 21V as i previously thought and is effectively pre-regulated. I wonder if Naim did this just so the 317 in the tray does not get too hot ?
Also i mis-traced the output from the second lm317 on the back panel. It also goes to the swing out tray via an RC filter and connects to the sensor (tells the player when the door has been closed) plus one or two other things i have not worked out yet.
The circuitry below the output relay still puzzles me.
Also the relay has an 0V connection right back to near the dac. Still dont understand why.
Stu
mega lord
Centre tapped
Hello John Luckins if you are reading.
Going back to your posts a couple of pages back about improving the ALWSR. I am ready to embark on these.
I have read and re-read your advice, looked at the circuit boards and the schematics a few times.
Initially i am going to do the tpr mods. Just to remind you, i have lm1086ct in that position not lm317t. You mention dropping R2 from 1K to 240R. That is no problem, i have some 240R now. You also mention swapping out R1 for a LED. I have looked and re-looked at the circuit and i think you have typed R1 when surely it must be R3 (1K) that you meant rather than R1 (100R) ? I have read your advice on a Mr Tibbs thread and there you state it is R3 which is replaced by an LED.
Sorry to be a pain, just want to be sure.
I will be measuring the current draw of the loads and alwsr (separately) this coming week and will give you the circuit details of the loads to help with the more complicated mods.
Many thanks, Stu
Going back to your posts a couple of pages back about improving the ALWSR. I am ready to embark on these.
I have read and re-read your advice, looked at the circuit boards and the schematics a few times.
Initially i am going to do the tpr mods. Just to remind you, i have lm1086ct in that position not lm317t. You mention dropping R2 from 1K to 240R. That is no problem, i have some 240R now. You also mention swapping out R1 for a LED. I have looked and re-looked at the circuit and i think you have typed R1 when surely it must be R3 (1K) that you meant rather than R1 (100R) ? I have read your advice on a Mr Tibbs thread and there you state it is R3 which is replaced by an LED.
Sorry to be a pain, just want to be sure.
I will be measuring the current draw of the loads and alwsr (separately) this coming week and will give you the circuit details of the loads to help with the more complicated mods.
Many thanks, Stu
martin clark
pinko bodger
TPR for a super-reg is just the same as any other - so green LED is fine over red; dropping the R value no problem either (it's usually a good idea because it help ensure the pre-reg 'sees' enough minimum load to regulate, always)
martin clark
pinko bodger
For total clarity can you post a link to the diagram you're referring to?
mega lord
Centre tapped
Hello Martin,
Its this schematic i was referring to.
Thanks, Stu
Its this schematic i was referring to.
Thanks, Stu
I wouldn't replace a well sorted ALWSR with a TPR'd LM317 based regulator and I've built quite a few of both though not specifically the TPR4. However there are some quite significant and straightforward improvements that can be made to the original ALWSR circuit. I would recommend R-C filtering the supply to the Opamp (a resistor between OP2 and C3, making sure C3 is at least 22uF). This involves cutting a track and checking the current draw off the opamps data sheet so not too many volts are dropped.
I would also recommend splitting R4 into two series 5k resistors and filtering their middle to SRIN with a large electrolytic, say 47uF-100uF. This makes for a stiffer CCS.
The TPR on the ALWSR circuit can be improved along the lines of the Acoustica TPR mods by removing C9 altogether and replacing R1 with a red or green LED and reducing the value of R2 from 1k to circa 240 ohms.
Finally the ALWSR is more stable when C1 has high ESR, C7 is not fitted and C5 is fitted (after the Opamp supply is decoupled) with a high ESR capacitor, say greater than 10uF with 0.5 ohms ESR.
There are other mods you can do to the reference voltage and the current source if you have enough voltage headroom but these ones above will give you a taste of the improvements available. What are your ALWSR's input and output voltages and which Opamp has it got?
http://[URL=http://s869.photobucket.com/user/John_Luckins/media/ALWSR%20ve%20circuit_zpszwjgx97q.jpg.html][/URL]
You could do these mods while listening to the TPR4 in there and then swap the ALWSR back??
John
john.luckins
pfm Member
mega lord
Centre tapped
I managed to measure the current draw of the 3.5 anologue stage toda.
The alwsr supplying the output opamps (2x opa627 only) draws 15.7mA between them (about 8mA each). They are 2 pole active lowpass filters and also drive the output (see acoustica).
The alwsr draws 33.6mA so 33.6 + 15.7 = 49.3mA total
The other alwsr supplying the other 4 opamps (opa604) & the output relay draws 62.4mA. This is the anti alias filter believe.
The alwsr draws 32.7mA so 32.7 + 62.4 = 95.1mA total.
They are wired in exactly as the acoustica instructions so the original lm317's are removed, the alwsr take their power from the former lm317 in position and the output goes to the former 317 out pin. Obviously there is no adjust used but the rest of the 317 circuit is left in.
Before
After
Hope martin does not mind me using his pics
That means the alwsr output not only goes straight to the load but there is also a 10uf tant straight to ground (former lm317 output tant) and a 200R in series with a 10uf tant to ground (the tant having the former 317 Vset 2.7K ohm resistor across it.
I am not sure what effect this little lot is having but it cant be that bad or Martin would not have left them in on his acoustica instructions.
Is the output tant to ground providing some isolation from the very low esr Wima 0.1uf film which decouples right at the opamp supply pin? I had read alwsr did not like very low esr caps such as these.
What the 200r and 10uf tant to ground (tant having 2.7k across it ) is doing i do not know. Drawing a bit extra load ? Maybe this can be removed leaving just the 10uf tant to ground.
The alwsr supplying the output opamps (2x opa627 only) draws 15.7mA between them (about 8mA each). They are 2 pole active lowpass filters and also drive the output (see acoustica).
The alwsr draws 33.6mA so 33.6 + 15.7 = 49.3mA total
The other alwsr supplying the other 4 opamps (opa604) & the output relay draws 62.4mA. This is the anti alias filter believe.
The alwsr draws 32.7mA so 32.7 + 62.4 = 95.1mA total.
They are wired in exactly as the acoustica instructions so the original lm317's are removed, the alwsr take their power from the former lm317 in position and the output goes to the former 317 out pin. Obviously there is no adjust used but the rest of the 317 circuit is left in.
Before
After
Hope martin does not mind me using his pics
That means the alwsr output not only goes straight to the load but there is also a 10uf tant straight to ground (former lm317 output tant) and a 200R in series with a 10uf tant to ground (the tant having the former 317 Vset 2.7K ohm resistor across it.
I am not sure what effect this little lot is having but it cant be that bad or Martin would not have left them in on his acoustica instructions.
Is the output tant to ground providing some isolation from the very low esr Wima 0.1uf film which decouples right at the opamp supply pin? I had read alwsr did not like very low esr caps such as these.
What the 200r and 10uf tant to ground (tant having 2.7k across it ) is doing i do not know. Drawing a bit extra load ? Maybe this can be removed leaving just the 10uf tant to ground.
martin clark
pinko bodger
You can remove the 100R+10uF combo, but it does no harm -nor does it contribute anything (unless, as I did, you want to swap a 317 back in later just to double-check what is gained)
And you are right, the 10uF tant on the board at the output of the alwsr is what helps keep it stable even with the 0.1uF film bypasses at the point of load. It's enough for the 317 too, note Fast regs like these get squirrely when you are loading them only with small value, low-esr caps
The total current draw is very low as you have found, and that's why IMO it's not worth separating the regs further; alwsrs, in fact most fast /discrete regs actually perform better with a bit of load on, and c.8mA for one opamp definitely isn't it! 50mA or more , as here, is better.
ETA: I actually stopped using that player over 10yrs ago
And you are right, the 10uF tant on the board at the output of the alwsr is what helps keep it stable even with the 0.1uF film bypasses at the point of load. It's enough for the 317 too, note
Fast regs like these get squirrely when you are loading them only with small value, low-esr capsThe total current draw is very low as you have found, and that's why IMO it's not worth separating the regs further; alwsrs, in fact most fast /discrete regs actually perform better with a bit of load on, and c.8mA for one opamp definitely isn't it! 50mA or more , as here, is better.
ETA: I actually stopped using that player over 10yrs ago
mega lord
Centre tapped
10 years ? Bloody hell. I take it the end was a smoked decoder?
Thanks for the advice Martin.
Maybe the alwsr on the output opamps (15.7mA draw) could do with a resistor to ground to get the current draw up to 50mA ? More load on the psu (= noise) but would the trade off be worth it.
At 8mA maybe a couple of fleas would be better for the two output opamps ? At 20V output maybe that would be too much heat for the 797.
I have been mulling the anologue stage over a lot recently. Mr tibbs 0V split mod, star wiring the feeds to the separate opamps (but can't use remote sense for this arrangement), alwsr mods. It never ends!
Stu
Thanks for the advice Martin.
Maybe the alwsr on the output opamps (15.7mA draw) could do with a resistor to ground to get the current draw up to 50mA ? More load on the psu (= noise) but would the trade off be worth it.
At 8mA maybe a couple of fleas would be better for the two output opamps ? At 20V output maybe that would be too much heat for the 797.
I have been mulling the anologue stage over a lot recently. Mr tibbs 0V split mod, star wiring the feeds to the separate opamps (but can't use remote sense for this arrangement), alwsr mods. It never ends!
Stu
martin clark
pinko bodger
There's no point in going split-rail, it is topologically identical to a single supply rail and coupling caps, if you think about it. And have done mod #7.
And the CD3.5 analogue stage supply/decoupling network on the pcb is already a very, very refined design; you'll find the way supply decoupling ('noisy current') routes are segregated from signal 0v reference not going to be improved upon, practically
And the CD3.5 analogue stage supply/decoupling network on the pcb is already a very, very refined design; you'll find the way supply decoupling ('noisy current') routes are segregated from signal 0v reference not going to be improved upon, practically