I'll try the AD797 next. Ordered a few off eBay (?). Looks like an excellent upgrade part for oscillators. Thx-RNMarsh
I'm tearing into the ShibaSoku analyzer. Signal traced the monitor output to a card and opamp output. Cant find anything on the output IC -- an NEC C813c. Being feed thru ss switches from TL0x opamps... possibly the switched/selected noise filters from the front panel... they noticably increase noise level when switched in. Need schematics... will start replacing opamps like in HP-339A as it dropped noise floor a lot and these old model opamps could do same if right ones are replaced. Been looking... need schematics.... where can i find them and the NEC opamp?
Thx-RNMarsh
Thx-RNMarsh
Last edited:
If you want to tear into the Shibasoku we should use my older one. Don't risk the new one. The manual has not arrived?
The system digitizes in a discrete digitzer stage (board 4 or 5 from the left I think). The filters should all be downstream from the notched audio. The harmonic processor is pretty complex.
The box way predates widespread use of FFTs so they may not have been looking at noise the same way. The TL072 is not noisy but not real quiet. However the other circuit elements may not allow a better noise floor with a lower noise opamp. This is common in instruments designed to specs.
The system digitizes in a discrete digitzer stage (board 4 or 5 from the left I think). The filters should all be downstream from the notched audio. The harmonic processor is pretty complex.
The box way predates widespread use of FFTs so they may not have been looking at noise the same way. The TL072 is not noisy but not real quiet. However the other circuit elements may not allow a better noise floor with a lower noise opamp. This is common in instruments designed to specs.
I ordered it from a japenese language site.... bet it didnt get thru.
I'm not going to try to lower the inherant noise level of the analyzer.... way too complex. But the distortion monitor/residual output stages where the filters and output opamps are located to get lower noise floor of what they are putting out for a residual signal. Dropping the noise 6dB would be a big help in seeing the harmonics at < .0001% levels. Its real tempting to use FFT on residual output to average out the random noise but try to get best results from analyzer' as is' first. If I dont find schematics, I might then take you up on doing the older unit. Thx-RNMarsh
BTW - does your unit have the space on the mother board for the GPIB card? Slot number 1 next to power supply on right facing front.
I'm not going to try to lower the inherant noise level of the analyzer.... way too complex. But the distortion monitor/residual output stages where the filters and output opamps are located to get lower noise floor of what they are putting out for a residual signal. Dropping the noise 6dB would be a big help in seeing the harmonics at < .0001% levels. Its real tempting to use FFT on residual output to average out the random noise but try to get best results from analyzer' as is' first. If I dont find schematics, I might then take you up on doing the older unit. Thx-RNMarsh
BTW - does your unit have the space on the mother board for the GPIB card? Slot number 1 next to power supply on right facing front.
Last edited:
uPC813
j-fet input low-offset operational amplifier
"The uPC813 is a high speed version of the uPC811. NEC's unique high-speed PNP transistor(ft=300mhz) in the output stage yields a high slew rate of 25v/us under voltage follower conditions without an oscillation problem."
NEC LINEAR IC DB 1988 uPC811 cross ref to LF411 in same databook
j-fet input low-offset operational amplifier
"The uPC813 is a high speed version of the uPC811. NEC's unique high-speed PNP transistor(ft=300mhz) in the output stage yields a high slew rate of 25v/us under voltage follower conditions without an oscillation problem."
NEC LINEAR IC DB 1988 uPC811 cross ref to LF411 in same databook
Hi Rick,
As far as I know NEC hasn't been making op amps in quite a few years. Are they even around anymore? That looks like a house number you put up there.
Can you figure it out from the surrounding circuit? Or is it something special?
Should have read further. Never mind.
Last edited:
Did I mention the THd of the KH-4402B with 2-opamp upgrades and a couple tuning adjustments is a really nice .0001% at 10Hz and .00013% at 1Khz and at 10Khz is .00015% THD+N. at 1volt into high Z. ?
If you just need a nice, clean oscillator that is flexible in output and frequencies.... it does very well.
[With the oscillator opamp changed out (LT1486 was used). No instabilities or weirdness with the opamp change.]
Thx-RNMarsh
If you just need a nice, clean oscillator that is flexible in output and frequencies.... it does very well.
[With the oscillator opamp changed out (LT1486 was used). No instabilities or weirdness with the opamp change.]
Thx-RNMarsh
Last edited:
Several have, I have heard that it is twitchy. In any case I expect you would have to trim THD at every major frequency change.
I've built it using Frex's boards. Once you get the hang of it, and set it carefully, it starts up consistently. I'll go back and find my post of the FFT for it.
Ken
[3rd round] AN67 Ultra-low THD 10kHz sine oscillator PCB Group buy. - Discussion Page 10 - diyAudio
The measurement was taken using Dick's active twin T fundamental null filter, fed to an HP3562, screen capture via HPIB VEE. Dick and I have discussed wether the 2nd H is from the twin T or oscillator. You will need to ignore the THD calc.
I don't know if the AN67 would be happy being switched between different frequencies. It's twitchy to get started oscilating, once set it seems fine. My plan is to have a few of them each set for different frequencies.
Ken
The measurement was taken using Dick's active twin T fundamental null filter, fed to an HP3562, screen capture via HPIB VEE. Dick and I have discussed wether the 2nd H is from the twin T or oscillator. You will need to ignore the THD calc.
I don't know if the AN67 would be happy being switched between different frequencies. It's twitchy to get started oscilating, once set it seems fine. My plan is to have a few of them each set for different frequencies.
Ken
BTW - the thd+n does not change with load of 600 Ohms at any freqs.
The 2h is -122 and 3h is -124dB at 1v/1KHz into 600 Ohms. KH4402B noise floor is -130dB. 100KHz BW. [Ref 1KHz amplitude is -110/10dB/div]
Thx-RNMarsh
Last edited:
BTW - the thd+n does not change with load of 600 Ohms at any freqs.
The 2h is -122 and 3h is -124dB at 1v/1KHz into 600 Ohms. KH4402B noise floor is -130dB. 100KHz BW. [Ref 1KHz amplitude is -110/10dB/div]
View attachment 335449
Thx-RNMarsh
These are some impressive results especially at 10kHz. The LT1468 is nice little op amp.
What output level does the oscillator run at. Have you tried lowering/raising the osc level to study the effect?
The LT1468 has astonishing low distortion in an SVO when operated below 3Vrms and minimal load.
Cheers,
Last edited:
Seems I cant add numbers today.... Full Scale of -110db =1 v output from monitor port and another -22-24dB would be -132 and -134dB. It's noise floor is about -140 -145dB. Very close to my lower limit of testing. Got to work on that
I use that opamp in everything... including the osc in the 339A. It seems to be better than its specs. The KH4402B output is variable up to several volts. I havent measured its max output... what ever the specs say it is, no doubt. Hold on .... I get about 6.5v rms from the 50 ohm output into 600 ohms. At that level, the THD+N is a little lower (better S/N) at .00008%.
The stock buffer opamp/output circuit is still in place! Its my NOS favorite osc as the thd doesnt change with scale selection/range and level to any significant degree.
Of course the ShibaSoku oscillator has lower noise floor and thd but not by much any more. It's THD+N is still 10dB better.
Thx-RNMarsh
I use that opamp in everything... including the osc in the 339A. It seems to be better than its specs. The KH4402B output is variable up to several volts. I havent measured its max output... what ever the specs say it is, no doubt. Hold on .... I get about 6.5v rms from the 50 ohm output into 600 ohms. At that level, the THD+N is a little lower (better S/N) at .00008%.
The stock buffer opamp/output circuit is still in place! Its my NOS favorite osc as the thd doesnt change with scale selection/range and level to any significant degree.
Of course the ShibaSoku oscillator has lower noise floor and thd but not by much any more. It's THD+N is still 10dB better.
Thx-RNMarsh
Last edited:
Hi Rick,
I meant the oscillator not he output of the unit.
For example the 339a osc runs at 9Vpp. The output is variable by attenuation.
Oh I see. It runs high... +10 and -10 peak or 20v peak to peak on schematic. I didnt probe it to see exactly. Nor attempt to change the level... just tune for lowest thd.
Note the opamp has a discrete complementary output driver stage added onto it.
I think it is safe to say now that the Harmonic levels shown by Demian of the harmonics of his ShibaSoku generator is actually that of the ADC system.
The 2H and 3h are running at .00003% or less (noise) from the AD725D analyzer display reading. Right down there with Viktor's levels.... you can use his generator to gauge the ADC system accuracy.
-RNM
Note the opamp has a discrete complementary output driver stage added onto it.
I think it is safe to say now that the Harmonic levels shown by Demian of the harmonics of his ShibaSoku generator is actually that of the ADC system.
The 2H and 3h are running at .00003% or less (noise) from the AD725D analyzer display reading. Right down there with Viktor's levels.... you can use his generator to gauge the ADC system accuracy.
-RNM
Last edited:
I reported on this a few days back somewhere. The entire unit is different EXcept the choice of IC for the oscillator. That remained the same. I have one of each oscillator ($$) just to see what could be done for myself and for the DIY'er and when stock, the 4402 is 6 db lower across the board. It allowed the spec to be 1/2 of the 4400. But its still a lot of work/changes for 6dB. This is consistant with the work on the 339A... most improvement was found in changing the osc IC and lowering system noise thoughout [and bad quality parts]. I didnt go further, as davada did.
On the 4402, just osc and tune up parts.
-RNM
Last edited: