Many use their ears as measuring instruments.
Better to use a voltmeter as a measuring instrument.
A basic DMM that has an unknown frequency response is better than ears.
Set to 199.9mVac scale and you can measure down to 0.1mVac resolution.
If it reads 0.0mVac then that means that the average is less than 0.05mVac
and when it reads 0.1mVac then you know that your very inaccurate measurement is somewhere between 0.051mVac and 0.149mVac ±the tolerances specified in the user manual. There are usually two separate tolerances that need to be added together.
0.5% of the reading and one or two of the least significant digits. some will be lower than 0.1% of reading but at very low reading this becomes insignificant because the ±1digit is far wider.
A reading that says 1.1mVac would be ±0.5% (±0.0055mVac) and one digit would be ±0.1mVac giving a tolerance band of ±0.1055mVac.
That gets applied to the 1.1mVac to give a range of lower end: 1.051mVac -0.1055 = 0.9455mVac to upper end: 1.149mVac +0.1055 = 1.2545mVac
or ~0.9mVac to ~1.3mVac for a 1digit tolerance DMM
see what I meant by "your very inaccurate measurement". but it's way better than kneeling down in front of your speaker and declaring "it's silent"
Then there is the inaccuracy introduced by the unknown frequency response. I know that my most trusted handheld has a rising response from ~15kHz to ~40kHz and then plummets above ~60kHz
But it allows good comparisons, even at 50kHz.
For a headphone amplifier I expect at least 20dB better than that. They are enormously sensitive. That's why I suggested the +28dB amplifier to get a useful measurement of the added input noise from the pre-amp.
Have a look at SGK's pre-amp Thread. He is trying to measure a Hum+Noise of -119dBu (~0.8uVac after applying the audio band filters)
Better to use a voltmeter as a measuring instrument.
A basic DMM that has an unknown frequency response is better than ears.
Set to 199.9mVac scale and you can measure down to 0.1mVac resolution.
If it reads 0.0mVac then that means that the average is less than 0.05mVac
and when it reads 0.1mVac then you know that your very inaccurate measurement is somewhere between 0.051mVac and 0.149mVac ±the tolerances specified in the user manual. There are usually two separate tolerances that need to be added together.
0.5% of the reading and one or two of the least significant digits. some will be lower than 0.1% of reading but at very low reading this becomes insignificant because the ±1digit is far wider.
A reading that says 1.1mVac would be ±0.5% (±0.0055mVac) and one digit would be ±0.1mVac giving a tolerance band of ±0.1055mVac.
That gets applied to the 1.1mVac to give a range of lower end: 1.051mVac -0.1055 = 0.9455mVac to upper end: 1.149mVac +0.1055 = 1.2545mVac
or ~0.9mVac to ~1.3mVac for a 1digit tolerance DMM
see what I meant by "your very inaccurate measurement". but it's way better than kneeling down in front of your speaker and declaring "it's silent"
Then there is the inaccuracy introduced by the unknown frequency response. I know that my most trusted handheld has a rising response from ~15kHz to ~40kHz and then plummets above ~60kHz
But it allows good comparisons, even at 50kHz.
1mVac into normal speakers is an enormous Hum+Noise. I expect <<0.3mVac and strive for a reading of 0.0mVac.
For a headphone amplifier I expect at least 20dB better than that. They are enormously sensitive. That's why I suggested the +28dB amplifier to get a useful measurement of the added input noise from the pre-amp.
Have a look at SGK's pre-amp Thread. He is trying to measure a Hum+Noise of -119dBu (~0.8uVac after applying the audio band filters)
Last edited:
For measurement then should output be loaded with dummy resistive load? In case of headphones it's 60ohms. Ground input or short with say 1k resistor to simulate impedance of DAC driver?
I have not found the load to be needed.
But at testing I never connect a speaker.
I do all my tests and repeat them long before I even think about adding a speaker.
There is an exception.
I have an old 4ohms car speaker above my workbench.
It has a 1k resistor in line. I can use that to monitor tones when testing with signal inputs.
But at testing I never connect a speaker.
I do all my tests and repeat them long before I even think about adding a speaker.
There is an exception.
I have an old 4ohms car speaker above my workbench.
It has a 1k resistor in line. I can use that to monitor tones when testing with signal inputs.
Here is the measurement using Fluke 101 in mV AC mode. I get 0.1mV, not quite zero but pretty close.
I might try twisting those AC mains cables and AC outs from the trafo and see if that makes a difference. It may be more related to the PSU which has only CRCRC and not CRCLRC like my other (larger) PSU.
According to Fluke 101 manual, the 600mV range is 3% of reading +/-3 of least significant digit.
http://img.banggood.com/file/products/20160724213004101_____umeng0000.pdf
Attachments
Last edited:
0.1mVac is good for a power amplifier driving normal sensitivity speaker.
I don't have equipment that can usefully measure the much lower noise coming from a Headphone Amplifier. That makes it difficult to suggest what to expect as a reasonable H+N.
But I still think you should be looking for at least 20dB lower than from a power amplifier.
That takes you down to the few tens of µVac
Did you read SGK Thread?
I don't have equipment that can usefully measure the much lower noise coming from a Headphone Amplifier. That makes it difficult to suggest what to expect as a reasonable H+N.
But I still think you should be looking for at least 20dB lower than from a power amplifier.
That takes you down to the few tens of µVac
Did you read SGK Thread?
I twisted all the wires into and out of the trafo and re-measured the noise with the DMM and saw no improvement. However, noticed that noise went down when connected to a battery powered source (phone) vs my DAC.
Sources are far more noisy than amplifiers when you're in the <0.1mVac range. Disconnect the source and short the input.
I noticed that ZM doesn't do any twisting of sensitive signal leads - but rather uses thin filament like wire floating like fishing line - maybe to reduce capacitance? Here is an example from his M2:
Also, his leads to and from trafo and PSU are not twisted either:
Also, his leads to and from trafo and PSU are not twisted either:
No Noise
It turns out that the simple banana jack to alligator clip test leads that I was using with the Fluke was picking up noise (when we are trying to see sub 100uV it makes a difference).
I put a Pomona Banana to BNC connector on the Fluke, then used RG174 coax BNC cable to a Pomona BNC to alligator test clip lead.
With input disconnected, volume pot turned up to max, no load connected, I now get 0.0mV on the Fluke.
Listening to the headphones with volume pot turned to max and no input connected (I think the input impedance of the cross feed filter is 5.5k ohms with no connection), I hear absolutely nothing. Turning the pot fully from zero to max I hear nothing, no scratchiness, nothing. Seriously, dead as a doornail and without the red LEDs showing juice flowing, one would be in doubt that the amp was plugged in and working. I am totally happy with its noise (or lack of noise) performance.
The unit is ready to ship as a gift now.
It turns out that the simple banana jack to alligator clip test leads that I was using with the Fluke was picking up noise (when we are trying to see sub 100uV it makes a difference).
I put a Pomona Banana to BNC connector on the Fluke, then used RG174 coax BNC cable to a Pomona BNC to alligator test clip lead.
With input disconnected, volume pot turned up to max, no load connected, I now get 0.0mV on the Fluke.
Listening to the headphones with volume pot turned to max and no input connected (I think the input impedance of the cross feed filter is 5.5k ohms with no connection), I hear absolutely nothing. Turning the pot fully from zero to max I hear nothing, no scratchiness, nothing. Seriously, dead as a doornail and without the red LEDs showing juice flowing, one would be in doubt that the amp was plugged in and working. I am totally happy with its noise (or lack of noise) performance.
The unit is ready to ship as a gift now.
Attachments
He made a comment that the amp is dead silent. Have you listened to the amp that he made? If not then how can you imply he is lying? You can have your own opinion and you do not have to believe him. But please keep it to yourself.
Andrew, do twisted wires really make that much difference to you? It's almost as if you have the ability to search diyaudio for images of untwisted wires.
I'd be more concerned about the low bias, undersized transformer and lm78xx regulators. All concessions to get it into that case I suppose.
I've tried this amp with low bias (100ma) and high bias (250ma). Higher sounds better. And you need more transformer for that. And I've tried lm317/337 vs the cap multiplier. Cap multi sounds better. I have headphones from 30 to 55 ohms.
Also, id worry about oscillation if you don't have caps on the amp boards.
But it's a nice, small build. Someone will be happy with it if the offset doesn't drift too much. Mine is pretty stable but the offset changes when the ac goes off at night at work. About a 15mv swing. I don't have $2000 headphones so I don't worry about it.
I'd be more concerned about the low bias, undersized transformer and lm78xx regulators. All concessions to get it into that case I suppose.
I've tried this amp with low bias (100ma) and high bias (250ma). Higher sounds better. And you need more transformer for that. And I've tried lm317/337 vs the cap multiplier. Cap multi sounds better. I have headphones from 30 to 55 ohms.
Also, id worry about oscillation if you don't have caps on the amp boards.
But it's a nice, small build. Someone will be happy with it if the offset doesn't drift too much. Mine is pretty stable but the offset changes when the ac goes off at night at work. About a 15mv swing. I don't have $2000 headphones so I don't worry about it.
I have my cardboard box amp that has a 50VA trafo - maybe I will try taking that to 250mA? That's 7.5w per channel burned off right? I can make some cap multipliers for it as well and help to reduce the dissipation in the voltage regulators.
I listened to the amp at 150mA and did not notice any improvement over 100mA with 60ohm headphones. Maybe needs 250mA to make a big difference but that is quite warm.
I listened to the amp at 150mA and did not notice any improvement over 100mA with 60ohm headphones. Maybe needs 250mA to make a big difference but that is quite warm.
I think it depends on the phones. If you feel that it's thin, more bias helps with a little more drive and authority. A little rounder. More drive. I'd try it and see if you like it. It all depends on how much heatsink you have and how stable you can keep the DC offset. I tend to take things to the extreme.
Keep an eye on the offset...It seems to drift on mine w/ambient temp slightly. Hotter might actually keep it more stable as it will be less influenced by outside temps. Obviously, more heatsink helps with this. I keep a spare 1/4 stereo male jack around and plug it in to test the offset once in a while, that way you don't have to take off the lid.
Those Sony monitor phones (I have a pair at work) are what I would consider dark and round sounding. Headphones vary so much in sound...more bias may not be what you are looking for but in general there are no drawbacks (other than heat and wasted energy) to increased bias.
I had mine initially running 317/337 and 130ma. It sounded good...but I like 200-ish ma and the cap multiplier better. More drive and weight which is what I am looking for..I am a speaker guy and to tell you the truth, headphones never impress me. I have a friend w/K701, hifiman, Beyer 990's...I've tried them all. They all have compromises.
It's fun to play with but I will take speakers over headphones anyday. Maybe I need to get better phones but I primarily listen at work so closed back is mandatory.
Chassis temps are still quite low with 230ma, warm to the touch but not hot. But I am using the 2107 chassis. I am also using an 18V transformer so I am dumping quite a bit of energy in the regs. With a 25VA transformer you are limited to about 125ma I would think.
I use Grados, AKG 272 and AT A900X as my personal phones. Headphones and headphone amps are highly subjective as far as I can tell, so you may have a different impression. But bias is free so it's a pretty cost effective improvement.
Next (maybe this weekend) I will increase rails to 18 or 19v. I don't need the voltage swing with my headphones but I want to take this thing as far as it will go.
Keep an eye on the offset...It seems to drift on mine w/ambient temp slightly. Hotter might actually keep it more stable as it will be less influenced by outside temps. Obviously, more heatsink helps with this. I keep a spare 1/4 stereo male jack around and plug it in to test the offset once in a while, that way you don't have to take off the lid.
Those Sony monitor phones (I have a pair at work) are what I would consider dark and round sounding. Headphones vary so much in sound...more bias may not be what you are looking for but in general there are no drawbacks (other than heat and wasted energy) to increased bias.
I had mine initially running 317/337 and 130ma. It sounded good...but I like 200-ish ma and the cap multiplier better. More drive and weight which is what I am looking for..I am a speaker guy and to tell you the truth, headphones never impress me. I have a friend w/K701, hifiman, Beyer 990's...I've tried them all. They all have compromises.
It's fun to play with but I will take speakers over headphones anyday. Maybe I need to get better phones but I primarily listen at work so closed back is mandatory.
Chassis temps are still quite low with 230ma, warm to the touch but not hot. But I am using the 2107 chassis. I am also using an 18V transformer so I am dumping quite a bit of energy in the regs. With a 25VA transformer you are limited to about 125ma I would think.
I use Grados, AKG 272 and AT A900X as my personal phones. Headphones and headphone amps are highly subjective as far as I can tell, so you may have a different impression. But bias is free so it's a pretty cost effective improvement.
Next (maybe this weekend) I will increase rails to 18 or 19v. I don't need the voltage swing with my headphones but I want to take this thing as far as it will go.
Also, another thing I have compared is the MUR860 rectifier vs the MSR860. Not in the F5hp but in the Pass BA-3 front end which is a close cousin. I built my F5hp with MSR860 from the start.
The MSR diode was superior (this was with a Jung reg, not the cap multiplier) and I've started using it in everything I can. They are cheap and EVUL has made a really good recommendation in his build guide. More detail than MUR860, easier sounding and and more delicate but still the same amount of weight and drive. I like them.
One more thing you could try.
The MSR diode was superior (this was with a Jung reg, not the cap multiplier) and I've started using it in everything I can. They are cheap and EVUL has made a really good recommendation in his build guide. More detail than MUR860, easier sounding and and more delicate but still the same amount of weight and drive. I like them.
One more thing you could try.
I actually got 150mA with my 25VA trafo fairly easily but case was warm to the touch as I have no heatsinks. How does a rectifier diode affect the sound? It is like way upstream behind a CRCRC and a regulator. I will make cap multiplier on my next one - I know they eat ripple like no one's business - but I had no issues with ripple or hum in my output as measured. I need to get a spare 1/4in TRS plug to do offset measurements. Mine does drift but if I let it warmup for an hour - it is stable within 15mV. It does move with temperature though so helps to keep it on all the time.
I am thinking of Sennheiser HD600's one of these days. The MDRV6's don't sound dark to me - but I like neutral balance. I agree that my best speakers beat headphones anyday. Especially my Trynergy, or my transient perfect 2-ways.
I am thinking of Sennheiser HD600's one of these days. The MDRV6's don't sound dark to me - but I like neutral balance. I agree that my best speakers beat headphones anyday. Especially my Trynergy, or my transient perfect 2-ways.
Last edited:
if your transformer and rectifier ring, then that leaks through the amplifier to appear at the speaker.
You need to check your PSU is not ringing.
That where the Quasimodo Thread came from.
You need to check your PSU is not ringing.
That where the Quasimodo Thread came from.
You mean switching transients from the diode bridge causing ringing? I have 330pF plus 10R snubbers at the output of each rectifier bridge formed by four 1N4004 diodes. So is using the fancy MUR or MSR TO220 style diodes supposed to not have ringing? Would the ringing be audible? As I said this amp is dead silent.
What's the Quasimodo thread?
What's the Quasimodo thread?
Last edited:
For those of you in the USA...massdrop (a group buying site) has senn 6xx on sale right now for $199 shipped. You have to sign up for the group buy but it's a particularly good deal. They regularly havr k701/k702 for $200 as well.