318's in mine, but I have no complaints:
An externally hosted image should be here but it was not working when we last tested it.
The range switching for the time being is a manual jump.
The switching will be done with relays controlled by a MCU and a small USB app.
I used Mdac's for tuning.
dick:
these improvements came from strictly dropping in a new op amp?
no changes to psu decoupling or anything else?
mlloyd1
Never mind - i see some of your comments later on some things I was curious about
btw, thanks to ALL you guys for sharing this stuff. great learning; just wish i had more time to play ...
these improvements came from strictly dropping in a new op amp?
no changes to psu decoupling or anything else?
mlloyd1
Never mind - i see some of your comments later on some things I was curious about
btw, thanks to ALL you guys for sharing this stuff. great learning; just wish i had more time to play ...
Last edited:
Seems exceptional for a 318 opamp. I'd look into how the instrument you are using which came up with that THD+N number. Never-the-Less, the numbers were better when replaced with a quieter and lower thd LT1468. My goal is to be as low as it is possible to be ... not just good enough stock. In some cases, I can tune for a single freq to be 10X better. add another zero. But measuring this accurately has been a side trip and challenge.
So what hardware/software are you using?
Thx-RNMarsh
Last edited:
I took the measurement after having spent about an hour adjusting the 4402B -- and I also found that the optimal output voltage was between 2 and 4 volts. Importantly, I took it on a Saturday morning when there was no bother from air conditioning etc. I used RG-59 cables that were only 6 inches long too.
The measurement was taken with an Audio Precision SYS-2722.
davada -- I am really looking forward to a new design as I have tried several existing sources and tried to upgrade/modify them. A new and flexible oscillator source of the same general level as Viktors or the ShibaSoku (Ag16 -- which i cannot find), would be a great addition. So, I'm done with the mods... until I find schematic of AD725D. I now have a base line from what is commercially available to compare against.
When I see my A-P again (its still in Thailand), I'll do similar to its oscillator to see how much more can be done to it. Such as drop in an AD797 (if my early Model One doesnt already use it?).
Thx-RNMarsh
When I see my A-P again (its still in Thailand), I'll do similar to its oscillator to see how much more can be done to it. Such as drop in an AD797 (if my early Model One doesnt already use it?).
Thx-RNMarsh
Last edited:
A quick calculation indicated that -120 dB from 1V (120 dB THD+N) would require a source with an impedance below 300 Ohms and no internal noise sources. Another metric is that its around 700 nV RMS. A quick check would be the measure the noise level with the input of the instrument shorted to get its internal noise floor. Then figure at least 120 dB up from that for the signal level. Measure the noise with the target bandwidth.
Usually THD +N is measured in a narrower bandwidth because ultrasonic noise or 1/F noise tells little about the THD. In audible terms A weighting or even better CCIR/ARM will give better insight as to the relative intrusion of noise on the signal. Except for microphone and phono preamps the noise floor of audio systems is really a non issue today. Even Kool, CA has more ambient noise that a good line stage, amp and normal efficiency speakers. Here in the big city, forget even thinking about it.
THD however is deterministic and can be audible because it is deterministic and audible much lower level than random noise.
Usually THD +N is measured in a narrower bandwidth because ultrasonic noise or 1/F noise tells little about the THD. In audible terms A weighting or even better CCIR/ARM will give better insight as to the relative intrusion of noise on the signal. Except for microphone and phono preamps the noise floor of audio systems is really a non issue today. Even Kool, CA has more ambient noise that a good line stage, amp and normal efficiency speakers. Here in the big city, forget even thinking about it.
THD however is deterministic and can be audible because it is deterministic and audible much lower level than random noise.
I don't think I will ever replicate it again -- most of the time the 4402B is 0.00035 to 0.00067. This measurement was done under pretty quiet conditions -- the computer monitor was moved away a few feet, the fluorescent lights were off, no solder station etc., etc. The 400Hz and 22kHz filters were on.
WRT 100kHz bandwidth -- if you want to measure noise, measure noise. If you want to measure THD, measure the value of each harmonic.
I did spend one summer working in a room enclosed in brass screening -- when I asked my alma mater if I could purchase it, they said the room had been turned into an undergrad lab!
If it is truly a 5534 then you can usually drop in a LT 1468 and gain 6-10dB of improvement.
If we are talking about the SYS1 then the OSC has four 5534 including the multiplier.
Can't accommodate the two pole comp if going to 1468.
The output amplifier is discrete. It looks more like a power amp.
Is that 6-10dB of disto, noise or both?
Cheers,
Pay attention:
You touch on all the points I have been trying to make.
1. you can drop the the distortion by changing the opamp and retune.
2. I am improving BOTh thd and noise.
3. The results i am forwarding doesnt require a screen room or anything unusual.
-RNM
Last edited:
But I have not tried it on the A-P... as I said it is in Thailand (Bangkok) a.t.t.
The changes have been to 318 osc opamp used in KH440/4402 models and the opamp in the HP339a so far.
I mention the A-P as something to try next BUTnoted for others as a bench-mark --> the spec for A-P One for thd+n is .001% or less. The KH and even the 339a can do as well or better when changing opamp and retuning.
Thx-RNMarsh
-Thx RNMarsh
Last edited:
One other oscillator type I'm interested in is a Twin T.
I don't know of any commercially available ones.
Larry sent me an interesting paper from Bell labs on Twin T networks.
bstj-twintee-1.pdf
It's too big to post here but if anyone is interested I can email it.
It shows how to use equal caps for tuning and other ways to manipulate the transfer function.
I don't know of any commercially available ones.
Larry sent me an interesting paper from Bell labs on Twin T networks.
bstj-twintee-1.pdf
It's too big to post here but if anyone is interested I can email it.
It shows how to use equal caps for tuning and other ways to manipulate the transfer function.
Last edited:
I believe the newer ShibaSoku is a twin-T (?)
I would be interested... pls email me at rmarsh@calweb.com (it isnt a secret unless you got spam to send me)
I am finishing loading a pcb of a twin-t notch and need to add the filter C and R's... this would be good timing.
I have a high precision LCR instrument to match parts when/if needed.
-RNM
Last edited:
David:
What is the title of the paper and who is(are) the author(s)?
I am wondering if this might be in my stash of bstjs i kept when i left.
thanks,
mlloyd1
What is the title of the paper and who is(are) the author(s)?
I am wondering if this might be in my stash of bstjs i kept when i left.
thanks,
mlloyd1