Audeo says the D8 uses a better part than the LM2575 so its moot I guess.
Max input voltage to YJ is 32V with one cap change, 22V without.
Thks Audeo for coming in, that is great news！
Happen to drop order for some parts, can you advice what inductor to get is best for hiss noise removal? 10uH, 15uH or ? what Inductor-Sat-current?
1 point i dint understand, recall you mention the noise comes from D8, not from the main smps supply, so shouldnt inductor goes after +Vout of MP2451 instead bef at +Vin? also shld I have a Cap to go with it?
Last edited:
As I said in the other thread, the more inductance the better, provided that it can withstand some decent current (3A should be enough, I suppose). Space is scarce in that little D8 box so you wouldn't be able to cram a huge choke there anyway, so you better measure it all yourself and see how tight you are prepared to go.
Yes, the inductor goes precisely there - right into the input of that DC-DC converter. I cut the input lane and put the choke across the cut. And yes, it wouldn't be bad to slap an additional cap alongside the one already there, but again, it would be awfully crowded so I did not bother. The end result is good enough for me: now I have to stick my ear into the speaker to hear the hiss whereas before I could hear it from a meter apart.
The Q is a direct function of the losses in the circuit, so it depends primarily on the ESR of the inductor. The 1mH I simulated with has ESR=2.3ohm (SLF7045T-102) but its 10% down current rating is 140mA. So its fine for an array of opamps but nothing much more current hungry than that.
I just ran an LTSpice sim for this inductor with a 1000uF/16V Nichicon HZ (ESR = 10mohm) and there's just a smidgen under 90dB attenuation @ 52kHz. No sign of any peaking in fact its looking over-damped. Fiddling with the ESR the Q=0.7 point comes for an inductor ESR of ~1.4ohm.
With almost 90dB attenuation we're looking at a couple of uV of the 52kHz ripple making its way through in theory.
By the way, Apogee's datasheet on this chip says: "Note, digital and power ground planes must be combined under the IC". And, as far as I remember, in Alientek D8 they made a single huge ground plane all around the chip. So, perhaps, isolating output power MOSFET's ground pins from digital signal ground with some choke would further help reducing hiss. But that would take a lot of effort and I did not even think of trying to do that.
No, I'm talking about the L2/C2 in the schematic you showed. See my reply to Nano (just above) for the values. If 140mA isn't enough then you'll need to use either a physically bigger inductor (for example SLF12555-102 : then add 0.33R series R, this'll handle 340mA) or reduce the attenuation by selecting a smaller inductance value in the 7mm size.
Before you hiss-killers do any permanent alterations (or damage) to you gear, you should know this: inherent noise of STA326 is very much dependent on DC voltage it is powered with. When I plugged 12V DC instead of 28V into Alientek D8 its noise level went down accordingly. So, if you don't need deafening roars from your sound system maybe you'll be content with this simple solution. I feed 12V to my amp from my PC via a fat LC filter and it is still very loud for my room.
Thks for caution & sharing great experience.
Think you may have a point, i guess the lowering supply lowers powerstage voltage - making noise less observable. on the other hand if MP2451 v-reg is indeed cause of noisy output, could it affect SQ such as subtle details gone missing, especially at quiet passage of music?
Meanwhile adding D8 mod cost pennies, to my larger partlist order.
I do besides meanwell 32~36v smps, have the use of laptop 19V smps as well.
Will listen out on noise to see how bad for average 86dB sensitive 6ohm speakers.
Think you may have a point, i guess the lowering supply lowers powerstage voltage - making noise less observable. on the other hand if MP2451 v-reg is indeed cause of noisy output, could it affect SQ such as subtle details gone missing, especially at quiet passage of music?
Meanwhile adding D8 mod cost pennies, to my larger partlist order.
I do besides meanwell 32~36v smps, have the use of laptop 19V smps as well.
Will listen out on noise to see how bad for average 86dB sensitive 6ohm speakers.
I already modded one of mine - I'm using 30V from an LT1083 linear reg and a separate linear supply for the digital section and I've removed the switching regulator (LM2575). My amp is fully silent with no signal lock. All roads lead to Rome?
Thanks!
Thank you so much for running a sim - I think you've found the simplest way to mitigate the LM2575. I can cut the trace from it to the 3.3V reg and put an inductor in the path, with a cap after it and the ground wire leading back to the ground pin of the LM2575. Even 1R + 47uH in series and 820uF shunt will give -70dB. E.g. my 47uH has 0.1 ohm DCR and the cap is Nichicon PLE with 0.005 ohm ESR so adding 0.5R in series is all I think I need for stability.
I think this filter will drop the DC voltage by less than 200mV so it still leaves 1.5V for the dropout on the AMS1117. All good.
Thanks!
Thank you so much for running a sim - I think you've found the simplest way to mitigate the LM2575. I can cut the trace from it to the 3.3V reg and put an inductor in the path, with a cap after it and the ground wire leading back to the ground pin of the LM2575. Even 1R + 47uH in series and 820uF shunt will give -70dB. E.g. my 47uH has 0.1 ohm DCR and the cap is Nichicon PLE with 0.005 ohm ESR so adding 0.5R in series is all I think I need for stability.
I think this filter will drop the DC voltage by less than 200mV so it still leaves 1.5V for the dropout on the AMS1117. All good.
Thanks!
Last edited:
BTW I ran the datasheet filter in a calculator and it says -40dB for 20uH and 100uF, while the datasheet claims -20dB.
The AMS reg has 100uV noise and decent ripple rejection at low frequencies (far less at 50kHz no doubt but still "some") so its likely that this reg will then set the noise floor.
The AMS reg has 100uV noise and decent ripple rejection at low frequencies (far less at 50kHz no doubt but still "some") so its likely that this reg will then set the noise floor.
Does anyone know virtual ground circuits well? I'm curious about the effects of changing the electros to smaller values... ripple current, bass roll off, more dynamic...? The amp applies a first order high pass filter to the SE outputs, set to between 80 and 360Hz so deep bass is not a concern. The filter seems to suit 4 ohm speakers at high frequencies.
I found an easy way to add the extra LC filter without cutting a trace - there's a diode that isn't needed after the 220uH (should be but perhaps it's 22uH) inductor.
An externally hosted image should be here but it was not working when we last tested it.
I found an easy way to add the extra LC filter without cutting a trace - there's a diode that isn't needed after the 220uH (should be but perhaps it's 22uH) inductor.
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.
Last edited:
Hi,
Sorry to pop in with a side question. Since one year, I drive LXmini speakers (LXmini Store) with 2 STA326 based amps (FX-Audio D802). The set us is Music Server => microcontroller doinf USB Async input / DSP/ SPdif output => FX-Audio D802.
The setup delivers high transparency, dynamics, image but... something is still missing. The system is not as emotionally engaging as expected. Difficult to put words on what is missing, but you are not fully captivated by the music. Maybe some lack of flesh / warmth...
Could this be the STA326 ?
Any idea about possible causes or experiments to do would be welcomed.
Best regards,
JMF
Sorry to pop in with a side question. Since one year, I drive LXmini speakers (LXmini Store) with 2 STA326 based amps (FX-Audio D802). The set us is Music Server => microcontroller doinf USB Async input / DSP/ SPdif output => FX-Audio D802.
The setup delivers high transparency, dynamics, image but... something is still missing. The system is not as emotionally engaging as expected. Difficult to put words on what is missing, but you are not fully captivated by the music. Maybe some lack of flesh / warmth...
Could this be the STA326 ?
Any idea about possible causes or experiments to do would be welcomed.
Best regards,
JMF
I'd wait and listen to it first. Some folks have had hiss problems with D802 and D8, but I find no audible hiss in my systems from these, unless I ramp up the gain and stick my head against the speakers.
This hiss does not depend on gain unless it is set to absolute zero. Try to feed something very quiet (but non-zero) to your amp and set gain to +1 and you'll get that noise floor we are talking about. Or, if you have D8, you can select AUX as its input while nothing is connected to it and get virtually the same result (ADC adds very little noise of itself). When digital stream to STA326 contains zeroes it usually (unless configured otherwise) switches its outputs off by shorting all speakers to the ground and thus you won't hear anything at all.
I ran the LM2575 through WEBENCH for a 300mA load - the L1 inductor can be 1mH and there'll be max 15mV noise/ripple peak to peak with a 330uF 170mR cap.
I also found a Coilcraft PFL3215-333MEB 1206 size inductor, 33uH, DCR 1.92 ohms, rated at 300mA that can replace the diode on the same pads and make the second LC filter. It's stable into 1500uF 0.05R. Another can also go in place of the diode that feeds the digital section and form a low pass there too.
I reckon these together should fix the hiss. We'll see. Parts ordered!
An externally hosted image should be here but it was not working when we last tested it.
I also found a Coilcraft PFL3215-333MEB 1206 size inductor, 33uH, DCR 1.92 ohms, rated at 300mA that can replace the diode on the same pads and make the second LC filter. It's stable into 1500uF 0.05R. Another can also go in place of the diode that feeds the digital section and form a low pass there too.
I reckon these together should fix the hiss. We'll see. Parts ordered!
It could be that the output filter on the amp is designed for 4 ohm and your speakers are 8 ? This would give an sparkly/grainy/brittle edge to the sound but it wouldn't have an effect on the bass. That should be superb. If not, move the speakers or change rooms. ;-)
Yes I know there is a permanent noise floor irrespective of signal (although this is as you say muted completely during silences). But I can't hear that hiss from any vaguely near normal listening position so it is for me a complete irrelevance although for some folks and some systems it clearly is a bother.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.