Yeah, gotta admit being surprised to find such a low noise reg with such an old design...
one of the modern regs I've been looking at is the reg113, dmos
Linear Regulators - Single Channel LDO - REG113-3 - TI.com
but the recent adp150 seems to be better so I replaced the Tps79333 in my dac with an adp150. I can't say it made a big difference and it's impossible to do a-b testing to be sure, but it seems to be a bit better.
I also ordered a few LM723 to try out. Their datasheet gives some load/line specs.
Regulated Output Voltage 5V
Line Regulation (∆VIN= 3V) 0.5mV
Load Regulation (∆IL= 50 mA) 1.5mV
http://www.national.com/ds/LM/LM723.pdf
So it seems the claim of 86dB is wrong and more like 55dB.
I am mostly experimenting with this circuit :
I've been wondering if I could try a darlington as a pass transistor - any thoughts on this anyone ? I've just been using a 2n5401 and that'll pass enough current but the transistor gets to 70 deg C. The next pass transistor to try is a BUP40.
Last edited:
If you use spdif, then also worth doing is replacing the pulse transformer and the output RCA socket with a BNC. Here is a photo of the Murata Da101 - note the output "earth" leg (the one closest to the Blackgate cap) is NOT connected to the board and the output RCA has been replaced with a BNC that is isolated from the case.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
once again thanks alot Tom,
i saw in your old post you were using vanguard XO do you still use it till now? or replace it with Tentlab?
right now i also use the vanguard and just ordering the tentlab. will it improves?
mostly i use coax. and in the i plant do use i2s from 550 to TDA1541
temporary i still use Hifiman HM801 as DAC.
I also have try some SDHC, like Transcend SDHC32GB Class 2, Transcend SDHC 32GB Class 4, Toshiba SDHC 32GB Class 4 and Toshiba SDHC 32GB class 10 and the result the best quality is from Toshiba Class 4 at no 1, TRanscend Class 2 at second position, Transcend Class 4 at third position, Toshiba Class 10 last position. actuay Toshiba class 10 is very good in detail separation and soundstage but the the low freq is cut quite big
have you try also comparing the SD CArd?
Rgds
BUDY
i saw in your old post you were using vanguard XO do you still use it till now? or replace it with Tentlab?
right now i also use the vanguard and just ordering the tentlab. will it improves?
mostly i use coax. and in the i plant do use i2s from 550 to TDA1541
temporary i still use Hifiman HM801 as DAC.
I also have try some SDHC, like Transcend SDHC32GB Class 2, Transcend SDHC 32GB Class 4, Toshiba SDHC 32GB Class 4 and Toshiba SDHC 32GB class 10 and the result the best quality is from Toshiba Class 4 at no 1, TRanscend Class 2 at second position, Transcend Class 4 at third position, Toshiba Class 10 last position. actuay Toshiba class 10 is very good in detail separation and soundstage but the the low freq is cut quite big
have you try also comparing the SD CArd?
Rgds
BUDY
Darlingtons will just increase the voltage drop. Are usefull when the regulater curent is higher than 1A. The incorporated OpAmp is really simple, i did obtain "back then" better results replacing it with a compensated LM201/301A instead, using just the reference voltage from inside 723. BAd part was that minimum voltage was 10V. I saw use of LM324 because of the minimum 3V voltage.
Last edited:
I hadn't thought about comparing cards. I have only 3 - a class 2 16Gb Sandisk, an Extreme III 16Gb and an Adata class 10 32Gb. I'll give it a go and see how it turns out.
I have a Tent 24.576 in an upsampling CD player and I tried a Van. 0.3ppm instead and was quite surprised that they sounded very similar. I haven't tried both in the QLS so I don't know if that circuit will be the same. So if you are using the Vanguard 0.3ppm then I'm not sure you'll hear much of an upgrade with the Tentlabs ? But if you also improve the power to the tcxo then perhaps you might. I'll be very interested to hear your opinion.
I was also quite surprised how much difference the bnc + pulse transformer made. A very inexpensive and worthwhile change.
I've tried a few different caps and, no surprise, the BG FK is definitely the best as a main cap. These are no longer in production and elsewhere IanS has recommended Nichicon PLE/PLF instead. I've ordered some and will try them as upgrades from the oscons, if they fit. I've never heard much of a difference, if any, going from one grade of oscon to another.
cheers,
Tom
Well, I was wondering if the higher hfe would make any difference, and if the wider bandwidth would either ?
A better op amp is definitely a good idea. I'm on the fence though, as I'm trying to keep component count down, but I'll look into it. 723 regs are dirt cheap and the circuit above needs only two caps, 3 resistors and a transistor so it's a very viable alternative to a 7805, which is what was in the QLS originally.
The circuit above, using a 2n4898, is supposed to give
Regulated Output Voltage +5V
Line Regulation (∆VIN= 3V) 0.5 mV
Load Regulation (∆IL= 1A) 5 mV (not really important as the measured load is constant)
so to achieve the 2.5uV noise output, ∆VIN<15mV. This can be done with common components too :
An externally hosted image should be here but it was not working when we last tested it.
LM317 has sligtly better noise than 7805. The above circuit uses the first reg as an amplifier for feed-back of the ripple back to the second reg (via capacitor C3). Helps with DC heat rejection too, by splitting the voltage drop betwwen the two regs.
I don't think it does anything for the noise, the inherent noise cannot be minimized by feedback.
I don't think it does anything for the noise, the inherent noise cannot be minimized by feedback.
Last edited:
Yep, the LEDs at the base of the second 317 give a lower impedance path to gnd for noise so it's a bit better than a standard 317 circuit, and easily good enough to allow the 723 to achieve its specs. The output of the 723 feeds a cap multiplier so I should be getting very very low noise at the tcxo - too low to measure with the gear I have. I may need to build one of Tangent's noise amps if I'm going to make any more progress. I am thinking about trying a Wenzel noise shunt after the 723 because that should reach the noise floor of the transistor but I suspect my soldering skills/build quality is probably the noise floor.
More mods that workHi,
I have replaced the 3.3V regs with adp150 (9uV rms noise to 100Khz) and they sound slightly better than the tl1963, which were much better than the original ams1117. I am using a 723 reg with BUP40 pnp pass transistor as the 5V source and this fed from a 10V Sigma11. I've changed the caps to Nichicon PLF 470 16uV and also changed the input filter on the power socket so it is a metallised poly 330nF cap followed by a ferrite bead. This sounds very very good now. I also tried some other cards - the best ones are Sandisk Extreme III 16Gb and Adata class 10 32Gb.
The adp150 are tsot package - a challenge to fit ! Here are some pics.
The PLF (note that the second of the orange met. poly caps has now been removed - too much sibilance)
An externally hosted image should be here but it was not working when we last tested it.
The adp150 on its side for the output connection
An externally hosted image should be here but it was not working when we last tested it.
On the top side, you can see I bent the centre pin (gnd) away from the other two (in and on) to make it easier
An externally hosted image should be here but it was not working when we last tested it.
a wire to connect in and on and the pad on the pcb
An externally hosted image should be here but it was not working when we last tested it.
and the wire for ground
An externally hosted image should be here but it was not working when we last tested it.
We have come to the conclusion that when you want to change the display, the push buttons (external or different position), the regulators, caps, and you only want to use I2S and hence would disable the SPDIF out, remote clock near the DAC, you will end up only using the dsPIC chip from the QA550 and little else.
So we decided to make our own, based very roughly on the Ultimate Source 2 from Koon.
The QA550 gives you a quick start, but is not the final solution, for us at least.
Patrick
So we decided to make our own, based very roughly on the Ultimate Source 2 from Koon.
The QA550 gives you a quick start, but is not the final solution, for us at least.
Patrick
looks like you could use some flux, thats waaaay too much solder on everything. another good and simple regulator to try is LTC6655 ultra-low noise reference, feeding lme49990 buffered by lme49610, this can easily be used as a negative regulator by switching non-inverting to inverting feedback. this is good for up to 250ma with a little sink on the buffer and has incredibly fast transient response with very wide bandwidth. idea taken from the lme49600 app note and tweaked a bit
Last edited:
Hi EVUL,
In the device I have now, I have changed clock, caps, pulse transformer and regs so you'll forgive me if I don't reach the same conclusion ;-) A bigger display for the QLS was nice but it didn't have a backlight as advertised so I returned it. I don't miss it in fact. Designing my own player and getting the PCB developed etc does seem like a slightly bigger project ?
Anyway, have you done any blind testing between the two ? I'm very interested if it sounds better ! I am also curious how you coped with writing and re-writing firmware ? Is there a link to the project ?
In the device I have now, I have changed clock, caps, pulse transformer and regs so you'll forgive me if I don't reach the same conclusion ;-) A bigger display for the QLS was nice but it didn't have a backlight as advertised so I returned it. I don't miss it in fact. Designing my own player and getting the PCB developed etc does seem like a slightly bigger project ?
Anyway, have you done any blind testing between the two ? I'm very interested if it sounds better ! I am also curious how you coped with writing and re-writing firmware ? Is there a link to the project ?
Last edited:
Oh yeah, but it proves if I can do it with a fat tipped iron and 0.8mm solder and no flux, anyone can.... ;-)
I use a max6250 in a Sigma11 for most of my digital gear, but for this, I wanted to try cheap easy stuff.
Last edited:
You underestimated our capability.
We of course design our own hardware and software.
And the DAC at the receiving end of the I2S, including the XO, is part of the total design concept.
Important is, you are happy with what do do and what you have.
Patrick
Just a quick update as the player is now as good as I can get it.
Critical caps have been replaced with Nichicon LE 470uF 6.3V (Digikey stock them) and a few other LE or LF caps here and there.
The ams1117 regs on the main chip and CS spdif have been replaced with adp150 and several LE caps.
The tcxo was replaced with Vanguard 0.3ppm and this gets power through a potential divider (5.6K and 22K) feeding a BC550 gyrator (very much better than an adp150 here).
5V is from a 723 reg with a BUP41 pass transistor and that gets its juice from a sigma11 set for 10.8V. Better than a battery.
I2s output via a 74VHC04, and spdif via a dc blocking cap and a newava s22083 pulse trans (output ground leg lifted from pcb).
If only it came like this. It's so much better it's ridiculous. It'd be quite easy to redesign for everything except the sigma11 input.
The only upgrade I'd like now is a CCHD957 xo but they don't do it for 11.2896Mhz.
Critical caps have been replaced with Nichicon LE 470uF 6.3V (Digikey stock them) and a few other LE or LF caps here and there.
The ams1117 regs on the main chip and CS spdif have been replaced with adp150 and several LE caps.
The tcxo was replaced with Vanguard 0.3ppm and this gets power through a potential divider (5.6K and 22K) feeding a BC550 gyrator (very much better than an adp150 here).
5V is from a 723 reg with a BUP41 pass transistor and that gets its juice from a sigma11 set for 10.8V. Better than a battery.
I2s output via a 74VHC04, and spdif via a dc blocking cap and a newava s22083 pulse trans (output ground leg lifted from pcb).
If only it came like this. It's so much better it's ridiculous. It'd be quite easy to redesign for everything except the sigma11 input.
The only upgrade I'd like now is a CCHD957 xo but they don't do it for 11.2896Mhz.
Hi John
long time never chat. the hiface you sent over is superb.
now i am looking at other alternative source. from the quotes below. i just want know if you are running this qa550 from lifepo4 battery only? how many batteries are you using?
how long a battery last?
thanks in advance,
erwin
long time never chat. the hiface you sent over is superb.
now i am looking at other alternative source. from the quotes below. i just want know if you are running this qa550 from lifepo4 battery only? how many batteries are you using?
how long a battery last?
thanks in advance,
erwin
I don't know how John did it with just one 3.3v Battery and no external supply. The tcxo on the board has a different reg than the pic.
One other thing is important. The I2S lines are way to long. I grab them from
the three resistors next to the pic. The next best solution is to cover the traces with liquid silver which needs to be grounded on each end.
Klaus
Hello battery lovers,
Have you tried this :
The χ1 Battery Management Board
followed by a divider/gyrator to create 3.3V ? I'm sure this would be lower noise than the battery directly.
I built 3 x1, including one with a single lipo cell and mcp73861 and no balancing chip of course. I tried it with an adp150 to regulate the battery to 3.3V but it was a step backwards so I didn't bother to try it with the divider/gyrator as I figured batteries are not the lowest noise power source. Did I get this wrong ? Are lifepo4 hugely better than lipo ?
Have you tried this :
The χ1 Battery Management Board
followed by a divider/gyrator to create 3.3V ? I'm sure this would be lower noise than the battery directly.
I built 3 x1, including one with a single lipo cell and mcp73861 and no balancing chip of course. I tried it with an adp150 to regulate the battery to 3.3V but it was a step backwards so I didn't bother to try it with the divider/gyrator as I figured batteries are not the lowest noise power source. Did I get this wrong ? Are lifepo4 hugely better than lipo ?