syn08 said:
More details:
1. + 2. The small components around 62KHz are a) due a HDTV TV running somewhere in the house (most likely teenagers watching a late night show 😀) and b) the DLM60-10 SMPSs feeding the amp. The measuring methodology is identical with the one used for PGP (and detailed on my web site). Essentially is calibrating the signal chain (that is, determining the Amber gain, fundamental attenuation and residual distortions) then measuring the output spectrum, substracting the residual spectrum, then rescaling the result with the signal chain gain. I have checked the individual spectral components with my PAR 124A lock in amplifier and they match to 10% which I think it's pretty good.
3. The 19.5KHz mid component is due to the way the 19+20KHz signal is generated. Rather than mixing two 10KHz and 20KHz signals, the system 100% AM modulates a 19.5KHz carrier with a high purity 1KHz signal. The carrier is then supressed but some residual remains. The carrier residual creates some extra IMD products (slightly visible at mid point between the 1KHz multiple frequencies) but they can be neglected FWIW. The system filters out the fundamentals as well, they are calibrated in a separate pass.
4. The SMPTE spectra are much more cleaner than in PGP. Reason for the missing 120Hz and up even components is feeding the amp from the DLM60-10 SMPSs rather than from a mains transformer.
5. The rise and fall times (defined as the slope of the pulse responses) are symmetrical and around 150nS.
Working on finalizing the noise measurements, damping factor (BTW, much better than in PGB) and stability margins (just responses in various capacitive loads). I'll be back...
Edit: Just in case, if it was not clear enough: these are measurements on the complete amp, that is, the OPS plus the mixed (IC plus discretes) front end, as in the schematics posted on my web site. The total gain is set to about 28.
AndrewT said:
They were around 100k and about double in size; I changed them late last night to a smaller size and a little better resolution to compensate.
Re: DF
Very possible. In YAP is actually 0.6uH || 4.7ohm vs. 2.5uH || 10ohm in PGP.
I'll do some calculations of substracting the inductive reactances and the damping resistors as soon as I'll finish the measurements.
Edmond Stuart said:
Very possible. In YAP is actually 0.6uH || 4.7ohm vs. 2.5uH || 10ohm in PGP.
I'll do some calculations of substracting the inductive reactances and the damping resistors as soon as I'll finish the measurements.
dimitri said:
Yes Dimitri, you got it right. I got lately the Amber 5500 extension board which allows direct IMD measurements.
Bob Cordell said:
myhrrhleine said:
Edmond Stuart said:
Thank you Bob and myhrrhleine!
Thank you Edmond for so gracefully shadowing me! 😀 😀 😀
Correction
This paragraph should read:
Rather than mixing two 19KHz and 20KHz signals, the system 100% AM modulates a 19.5KHz carrier with a high purity 500Hz signal.
I should proofread before pressing submit...
syn08 said:
This paragraph should read:
Rather than mixing two 19KHz and 20KHz signals, the system 100% AM modulates a 19.5KHz carrier with a high purity 500Hz signal.
I should proofread before pressing submit...
Damping Factor
Much better than in PGP. As Edmond correctly noticed, this is because of the much smaller output coil (0.6uH instead of 2.5uH).
I'm not going to disclose this to the listening tests subjects, I'm curious if it will have an audible impact. Pesonally, I don't think so...
Much better than in PGP. As Edmond correctly noticed, this is because of the much smaller output coil (0.6uH instead of 2.5uH).
I'm not going to disclose this to the listening tests subjects, I'm curious if it will have an audible impact. Pesonally, I don't think so...
An externally hosted image should be here but it was not working when we last tested it.
Re: Damping Factor
Don't worry, only the GEB will pretend it can hear the difference.
syn08 said:
Don't worry, only the GEB will pretend it can hear the difference.
Re: Re: Damping Factor
BTW, I've settled for this:
http://cp.literature.agilent.com/litweb/pdf/5965-5427E.pdf
I'll pick it up on monday. Has options AY7 (second channel) and UFG (extra 4 megs of memory). A little bit more expensive than I was expecting, but well worth I guess. It's supported by HP until 2010.
Edmond Stuart said:
BTW, I've settled for this:
http://cp.literature.agilent.com/litweb/pdf/5965-5427E.pdf
I'll pick it up on monday. Has options AY7 (second channel) and UFG (extra 4 megs of memory). A little bit more expensive than I was expecting, but well worth I guess. It's supported by HP until 2010.
THD+N vs. power
Here's the YAP THD+Noise vs. power. at 20KHz.
First, it is obvious that YAP is much noisier than the PGP; at low power, the noise is dominating the results and the equivalent is several time larger than in PGP. Measurements are not yet completed (I am waiting for the new equipment to arrive) but preliminary the unweighted equivalent input noise is around -111dB (-120dB in PGP). It is believed this (larger noise) is an unavoidable effect of the current feedback configuration, and because the relative large currents employed in the OPS and front end gain stages. I don't think it really matters, but I'll further look into this noise thing, by independently measuring the noise in the OPS.
Secondly, the nice surprise is that even at 150mA bias, there is a region where YAP crosses the 0.0001% (1ppm) barrier. This is happening before the half power (and maximum dissipation) is reached.
Third, the THD+N increase as approaching clipping is much softer than in the PGP amp. THD+N starts to rise shortly after half power.
Fourth, YAP has considerably more muscle than PGP. It is able to put 218W at 0.0002% (2ppm) and 260W at 0.004% (40ppm) THD20, both in a 4ohm load. I raised the supply at +/-60V and YAP put 340W in 4ohm with 0.005% (50ppm) THD20. I also did a quick check at 2ohm and YAP put 450W at 0.005% (50ppm) THD20, which I think it's quite remarkable.
I checked several times, and THD at 1KHz was always under my measurements capability.
Sometimes in the weekend, THD vs. frequency will follow.
Here's the YAP THD+Noise vs. power. at 20KHz.
First, it is obvious that YAP is much noisier than the PGP; at low power, the noise is dominating the results and the equivalent is several time larger than in PGP. Measurements are not yet completed (I am waiting for the new equipment to arrive) but preliminary the unweighted equivalent input noise is around -111dB (-120dB in PGP). It is believed this (larger noise) is an unavoidable effect of the current feedback configuration, and because the relative large currents employed in the OPS and front end gain stages. I don't think it really matters, but I'll further look into this noise thing, by independently measuring the noise in the OPS.
Secondly, the nice surprise is that even at 150mA bias, there is a region where YAP crosses the 0.0001% (1ppm) barrier. This is happening before the half power (and maximum dissipation) is reached.
Third, the THD+N increase as approaching clipping is much softer than in the PGP amp. THD+N starts to rise shortly after half power.
Fourth, YAP has considerably more muscle than PGP. It is able to put 218W at 0.0002% (2ppm) and 260W at 0.004% (40ppm) THD20, both in a 4ohm load. I raised the supply at +/-60V and YAP put 340W in 4ohm with 0.005% (50ppm) THD20. I also did a quick check at 2ohm and YAP put 450W at 0.005% (50ppm) THD20, which I think it's quite remarkable.
I checked several times, and THD at 1KHz was always under my measurements capability.
Sometimes in the weekend, THD vs. frequency will follow.
syn08
for being into 4 ohms this is one excellent performance!
not many DIY amplifiers can ever come close to this
Question:
Is this graph from one real test, with Dummy 4 Ohm resistor?
... if so, it is even better, than if 'only spice'
Now, even one plain SPICE Run,
especially when using Non-Ideal voltage supply = Model of Transformer + Rcifier diodes + Caps Filter,
can be a good hint of quality of amplifier
regards
Lineup
for being into 4 ohms this is one excellent performance!
not many DIY amplifiers can ever come close to this
Question:
Is this graph from one real test, with Dummy 4 Ohm resistor?
... if so, it is even better, than if 'only spice'
Now, even one plain SPICE Run,
especially when using Non-Ideal voltage supply = Model of Transformer + Rcifier diodes + Caps Filter,
can be a good hint of quality of amplifier
regards
Lineup
Re: Re: Re: Damping Factor
Lucky you! This is one great piece of equipment. Looking forward to your findings.
Jan Didden
syn08 said:
Lucky you! This is one great piece of equipment. Looking forward to your findings.
Jan Didden
Re: Re: Re: Re: Damping Factor
I was originally looking at a 8751A (lost a couple of EBay auctions) then at a 4195A (always to expensive, don't really understand why, unless you pick one from the far east, which I'm not keen to, risks are to high) and settled for the 89410A from a local surplus warehouse (no outrageous shipping charges from the US). Does both spectra and vector network analysis up to 10MHz which I think would be enough for audio open loop measurements. 500MHz would be an overkill.
janneman said:
I was originally looking at a 8751A (lost a couple of EBay auctions) then at a 4195A (always to expensive, don't really understand why, unless you pick one from the far east, which I'm not keen to, risks are to high) and settled for the 89410A from a local surplus warehouse (no outrageous shipping charges from the US). Does both spectra and vector network analysis up to 10MHz which I think would be enough for audio open loop measurements. 500MHz would be an overkill.
- Status
- Not open for further replies.