When I have some data. I already got to -87dB using 2N3055 and a LM317 reference with 47uF noise filter ( usually 10 uF in data sheets ). The idea of the NPN and PNP compound pair is same voltage loss with much higher current gain. It should be a resistor with current gain in concept. When time allows I will see if it can work. This compound pair makes people very cautious usually as they fear it will go unstable. As there is no loop feedback to the LM317 ( Or a TL431 if prefered ) I hope for good things.
It starts.
Next days I will test a lot of PO-3 transes. P-channel, pnp; mosfets: SK50 (Hitachi and an other manufacturer), IRF9240 (and a IRFP9240 TO-247 too); silizium: BD 318, 2N2955, MJ15004; germanium: 2N268. Finally I want build a SE with the 2N268;-)))) The same psu for all, the same bias, same input and output and source and speaker and so on. NO multiway speakers!
O.K.: Not the 2N3055 and MJ15003 I will test. Just the complementary transes. These DO NOT sound like the n-channel, npn! The biggest problem of complementary-parts-pp-amps! NOT the crossover distortions, but the audible differents of + and - of a signal/wave;-)
THD and TND are not relevant. I will tell the audible differences, characters!
And, I did not forget, the JLH with SOT-32 types I will build and test, and tell you about;-)))
Next days I will test a lot of PO-3 transes. P-channel, pnp; mosfets: SK50 (Hitachi and an other manufacturer), IRF9240 (and a IRFP9240 TO-247 too); silizium: BD 318, 2N2955, MJ15004; germanium: 2N268. Finally I want build a SE with the 2N268;-)))) The same psu for all, the same bias, same input and output and source and speaker and so on. NO multiway speakers!
O.K.: Not the 2N3055 and MJ15003 I will test. Just the complementary transes. These DO NOT sound like the n-channel, npn! The biggest problem of complementary-parts-pp-amps! NOT the crossover distortions, but the audible differents of + and - of a signal/wave;-)
THD and TND are not relevant. I will tell the audible differences, characters!
And, I did not forget, the JLH with SOT-32 types I will build and test, and tell you about;-)))
Let us know how it sounds. Very good luck. In some ways all THD tests are for is being sure the design is working the same sample to sample. Oscillation is nasty and worth looking for.
I did some more oscillator work today. Not very satisfied except I found a good cheap version using TL074/84, I will write it up some time.
I did some more oscillator work today. Not very satisfied except I found a good cheap version using TL074/84, I will write it up some time.
;-)))
It does run;-)
I start with the TO-247 IRFP9240. The circuit is quasi the same like a "matching"-circuit, or this one: De-Lite Amplifier
At moment I have to accustom to the "distortions", the "noise" of the parts: resistors, caps, wires...
Tomorrow the substitute starts.
But, of course, it sounds much much better than every purchasable amp;-)))
It does run;-)
I start with the TO-247 IRFP9240. The circuit is quasi the same like a "matching"-circuit, or this one: De-Lite Amplifier
At moment I have to accustom to the "distortions", the "noise" of the parts: resistors, caps, wires...
Tomorrow the substitute starts.
But, of course, it sounds much much better than every purchasable amp;-)))
YouTube
;-)))
Welcome (back) in the TO-3 -cosmos;-)
I did build in the TO-3 without any rack. Cutted the contact and set the wire. NEVER use screws and discs for circuit, for current.
Quasi the same parameters, IRF9240 and IRFP9240, but a big audible difference.-)
We talking about materials and diameters, regarding circuit and current, now;-!!!
More flat and more sluggish. Minimal overcast and indistinct. The depths a half tone behind. The highs smore sibilant. Less 3-D, less contours. More dreamy. Something like this.
The generally TO-3 character - compared with more little casings;-!!! (Again and again and again the same experience;-)
;-)))
Welcome (back) in the TO-3 -cosmos;-)
I did build in the TO-3 without any rack. Cutted the contact and set the wire. NEVER use screws and discs for circuit, for current.
Quasi the same parameters, IRF9240 and IRFP9240, but a big audible difference.-)
We talking about materials and diameters, regarding circuit and current, now;-!!!
More flat and more sluggish. Minimal overcast and indistinct. The depths a half tone behind. The highs smore sibilant. Less 3-D, less contours. More dreamy. Something like this.
The generally TO-3 character - compared with more little casings;-!!! (Again and again and again the same experience;-)
Attachments
![IMG_20180115_124342[1].jpg](/community/data/attachments/611/611729-01ad75ae9bcbcbcc6136abc957c5c68c.jpg)
![IMG_20180115_132452[1].jpg](/community/data/attachments/611/611746-7b1ed9b2c9852e96af37d11d64ca1e17.jpg)
YouTube
Now the 2SJ50 Hitachi. Compared with the TO-3 IRF9240.
Quieter;-) Finer the mids. Finer colors minimal. Minimal higher solution. Less depths. No better 3D! Not more contoured. Flat like a wallpaper too;-) Sluggish, overcast, indistict like the IRF9240.
Wallpaper-listening at moment;-)))
My mind: The most think, TO-3 sound good, but the unaccuracy does mask a lot of other problems, unaccuracies, like wirewound resistors, pcb, complementary-pp ...-)
Now the 2SJ50 Hitachi. Compared with the TO-3 IRF9240.
Quieter;-) Finer the mids. Finer colors minimal. Minimal higher solution. Less depths. No better 3D! Not more contoured. Flat like a wallpaper too;-) Sluggish, overcast, indistict like the IRF9240.
Wallpaper-listening at moment;-)))
My mind: The most think, TO-3 sound good, but the unaccuracy does mask a lot of other problems, unaccuracies, like wirewound resistors, pcb, complementary-pp ...-)
I have heard about these T03 IRF types. Very interesting.
I have tried many versions of the this 1981 John Lindsley Hood dual op amp oscillator and found my first version the best. Not perfect, but very low cost. Use 16K for 1 kHz. Batteries are fine and Zinc types often are a bit higher in voltage when new. These cost 4=£1`from CPC as a box. Batteries are ideal as no hum. Ths oscillator is good enough do basic tests. Interestingly if the purple output clips you don't always know at the green output. Best to check. 430R replaced by 300R will allow TL082( 72 ) to be used and offers almost identical performance at a slightly reduced output. I doubt other who who say they get - 90 dB. Anyone want to simulate it? Take the lamp to be 1/2 of 430R.
I did try 2 x NE5532 as greater current version. It was the same as the data suggests it would be. I did try a RA53 thermistor and found it similar. They are now very expensive ( always were ). The lamps as far as I know is the most common type ( 28V 40 mA ). Maplin UK stock 28V 24 mA type BT44X which I suspect is better ( try 680R ). From what I know 28V 40 mA is universal aircraft type. I guess run at 24V typical.
I have tried many versions of the this 1981 John Lindsley Hood dual op amp oscillator and found my first version the best. Not perfect, but very low cost. Use 16K for 1 kHz. Batteries are fine and Zinc types often are a bit higher in voltage when new. These cost 4=£1`from CPC as a box. Batteries are ideal as no hum. Ths oscillator is good enough do basic tests. Interestingly if the purple output clips you don't always know at the green output. Best to check. 430R replaced by 300R will allow TL082( 72 ) to be used and offers almost identical performance at a slightly reduced output. I doubt other who who say they get - 90 dB. Anyone want to simulate it? Take the lamp to be 1/2 of 430R.
I did try 2 x NE5532 as greater current version. It was the same as the data suggests it would be. I did try a RA53 thermistor and found it similar. They are now very expensive ( always were ). The lamps as far as I know is the most common type ( 28V 40 mA ). Maplin UK stock 28V 24 mA type BT44X which I suspect is better ( try 680R ). From what I know 28V 40 mA is universal aircraft type. I guess run at 24V typical.
YouTube
No pleasure with MosFets. I changed to bipolars now. BD318. And the sun rises;-)
Much much much more clean and much much more power and much much more silence. Louder too. Start into 3D! Into contour! No noise, no "din"-!!! A dream!!! Blackness. Not grey and gray;-)
This is my/an again-and-again-experience: NO MosFets. No chance to get a clean sound. Horrible. Nothing for home-audio! The TO-3 bipolar sounds much better than the/a TO-247 MosFet.
Will listen now 1 - 2 days, before I replace with 2N2955 and with the germanium;-)))
No pleasure with MosFets. I changed to bipolars now. BD318. And the sun rises;-)
Much much much more clean and much much more power and much much more silence. Louder too. Start into 3D! Into contour! No noise, no "din"-!!! A dream!!! Blackness. Not grey and gray;-)
This is my/an again-and-again-experience: NO MosFets. No chance to get a clean sound. Horrible. Nothing for home-audio! The TO-3 bipolar sounds much better than the/a TO-247 MosFet.
Will listen now 1 - 2 days, before I replace with 2N2955 and with the germanium;-)))
Which version of JLH are you testing?
Do you plan to include 2SC5200 or MJL3821A? (But check for oscillation and if it does try a small (33pF) capacitor across the feedback resistor.
Why not a 2N3055 (epi version now it sh only type available)?
From what I saw on the D/S of the BD318 it seems to be a pro-electron variant of their MJ802-MJ4502 type devices. Expect 2MHz fT (same as 2N3055 epi), not different from 2N3055 (epi).
Thanks
Do you plan to include 2SC5200 or MJL3821A? (But check for oscillation and if it does try a small (33pF) capacitor across the feedback resistor.
Why not a 2N3055 (epi version now it sh only type available)?
From what I saw on the D/S of the BD318 it seems to be a pro-electron variant of their MJ802-MJ4502 type devices. Expect 2MHz fT (same as 2N3055 epi), not different from 2N3055 (epi).
Thanks
Nigel- I have also wrestled with filament lamp oscillators. None work really well; distortion is always 0.01 to 0.1% (increasing with lower frequencies) due to horrid third harmonic because of the lamp filament resistance. Also problems with amplitude bounce. But on the other hand distortion at 10kHz can be quite low (0.01% or thereabouts).
I have found that the Wien bridge circuit with a "damped" filament lamp feedback (i.e. split feedback path where the lamp voltage is attenuated) gives reasonable damping and lower distortion than directly running the feedback from the resistor/lamp junction. My best design uses several op-amps with LED/LDR stabilisation as R(A)53's are now made from unobtainium.
Perhaps we should have another thread on oscillator design?
I have found that the Wien bridge circuit with a "damped" filament lamp feedback (i.e. split feedback path where the lamp voltage is attenuated) gives reasonable damping and lower distortion than directly running the feedback from the resistor/lamp junction. My best design uses several op-amps with LED/LDR stabilisation as R(A)53's are now made from unobtainium.
Perhaps we should have another thread on oscillator design?
YouTube
TO-3-pause;-) Short;-)
Changed the BD318 - wonderful TO-3 transe - with a TIP2955 in a TO-247 case.-)))
Better contoured, better 3D, all better on point. Better tempo, rhythm. Better solution. More natural glance.
BUT: Starts, to show the problems of other parts. At moment I do use - slaughtered the Hitachi HMK 7500 II for this study - wirewound resistors and a unregulated psu and a mini-capacitance: chain of 8 x 4 mF and 7 x 0,47 Ohm. And cinch-plugs and so on.
TO-3-pause;-) Short;-)
Changed the BD318 - wonderful TO-3 transe - with a TIP2955 in a TO-247 case.-)))
Better contoured, better 3D, all better on point. Better tempo, rhythm. Better solution. More natural glance.
BUT: Starts, to show the problems of other parts. At moment I do use - slaughtered the Hitachi HMK 7500 II for this study - wirewound resistors and a unregulated psu and a mini-capacitance: chain of 8 x 4 mF and 7 x 0,47 Ohm. And cinch-plugs and so on.
YouTube
But, how does the germanium 2N268 sound;-?!? It is in. And I compare with the BD318.
A bit quieter. Not black, but brown the colour is;-) Contoured, fine, 3D too. The depths left the contours minimal. Big background. Not as on a point as the BD318. A little bit dreamy. Minimal slowler, minimal less attack. Minimal dough-like. NO noise - somebody does apprehend;-?
In this configuration great;-!!! The most would cry - and ditch their 100.000€$£ pre-power-combi;-)))
Arabica and Robusta;-?!?
But, how does the germanium 2N268 sound;-?!? It is in. And I compare with the BD318.
A bit quieter. Not black, but brown the colour is;-) Contoured, fine, 3D too. The depths left the contours minimal. Big background. Not as on a point as the BD318. A little bit dreamy. Minimal slowler, minimal less attack. Minimal dough-like. NO noise - somebody does apprehend;-?
In this configuration great;-!!! The most would cry - and ditch their 100.000€$£ pre-power-combi;-)))
Arabica and Robusta;-?!?
My feeling is it's my duty to do a low cost oscillator so people can advance to doing less guessing. I have been told most sound cards will with software do Fourier analysis. You can build my idea for less than £5 including batteries . It will be good enough to test most things.
Any circuit John? I did try a similar idea to yours which wasn't clearly better. The thing I see is lots of > 40 kHz distortion if 10 kHz. Not really a problem as it is -60 dB. Not nice all the same. If you use 16K +10nF my Wien gives supurb 1 kHz. For LF use 10 nF + 318K for circa 50Hz.
For PNP version just reverse my BC327/337 ( 337 input and 327-40 driver). One could use BD140 driver. I would stick with 2N2955.
Me too. I have all the parts and will use the LM317+2N3055 PSU. It's like I am on the starting blocks for a big race. It will be my phone, JLH, Mission 760's when I do. And first track will be Laurie Anderson " One Superman ". If it works well I will borrow Krell CD player and BBC LS5/9 at my friends house. In my mind I will compare with Willaimson amps, Leak TL12+, Marantz 9 and Quad 303. Janine has a Leak Stereo 20 which I am certain will win, her Krell I expect to be beaten. She and I prefer the Leak. She has the Krell as her do anything amp. I like my phone more than my CD player and it has a volume control which saves building something. The amp will be made of 90% junk. I still expect it to beat the Krell.
Nigel - the only oscillator circuit I have tried using a filament lamp giving low distortion is to use it in a positive feedback arm of a twin-t oscillator. There are some issues, such as the distortion is dependent on the open loop gain of the oscillator (which gives very good results at lower frequencies) but many op amps need buffering to drive the lamp (your 28V proposal may be better than my 6V 60mA)) and that the opamp needs a resistor in series with the TT network (100ohms is about OK) and a small compensation capacitor (5..10pF) which affects OLG (bandwidth) and hence feedback requirements. THe feedback is typically the lamp+resistor (R=22 ohms) and then a 10k feedback (to + input) with a 100 ohm pot to adjust the feedback. Suspect it will have similar performance to your circuit but not better. Last time I checked the distortion was about 0.1% at VLF (a few cycles) and fell to 0.01% around 100Hz, rising again above 10kHz as the OLG falls off. But the gain needed to be adjusted for each TT network.
Last edited: