Jfet pre

[Emmathom]

Thank you so much @thor.zmt for these informations.
Yes, Thet from https://www.halfshavedyaks.xyz/index.php told me the same about Hampton's schematic : not involded at all...
the idea is not to try to make a transistor sound like a tube, but to keep this 4-channel preamp in a "mixed" version: 2 tube channels + 2 transistor channels (no AOP). This offers more possibilities, different colors in recording session.

I could very well have changed the HT transformer and added 2 tube channels but I want to keep this "mixed" configuration.

Since I do not want to work with OPA (I already have a quad OPA134 in gain + NE5532 in balance) because it lacks "soul" and the current circuit based on BJT does not offer any personality ( !) I thought of jfet.

So a few questions please :
Should i try the last schematic you posted ? (screen-capture attached)
How much gain does it offer ? (knowing that my OEP X187B 1:10 offers +20dB...)
What voltage is needed for this circuit ?
Does not the output R to ground to low ? 10KΩ // 10KΩ (Line In sound card)

 

[Emmathom]

Our replies have crossed...
So better try this schematic...

 

[thor.zmt]

So a few questions please :
Should i try the last schematic you posted ? (screen-capture attached)

You can. It should work ok in real life. But It is still a Hampton. Same gain, same powersupply voltage etc.

Look at my next post for something a bit more Juicy.

Does not the output R to ground to low ? 10KΩ // 10KΩ (Line In sound card)

Yes, I am actually simulate the 10K load of the sound card.

The schematics I post (unless hand drawn on the back of a napkin which happens a lot) cone from a software called "TINA" in the free version from TI.

This software is basically a virtual circuit and virtual instruments and allows you to at least get a fair idea of what the performance potential of a circuit is. Real circuits are always a little different, but overall this is pretty close to real.

Thor

 

[thor.zmt]

Our replies have crossed...
So better try this schematic...

Well, IT DEPENDS.

You might want a very colourful Mic Pre. Then Hampton, with "sonically colourful" passive parts like Allen Bradly or Xicon Carbon composition resistors, Mylar or polycarbonate capacitors and coloured electrolytic types like Axial Nichicon VX might be the way to go. Add very "Iron sound" input transformers (tiny) and a small size low nickel core output transformer.

As said, sonically it would be a caricature of a good tube circuit, everything overly emphasised, kinda like KFC "secret spice" (a truckload of MSG and deep/pressure fried in industrial oil) vs good southern fried chicken without MSG, Trans fats etc. et al....



I am not saying that a "colour preamp" is not valid, it is valid. Recording is Art, not science. Nobody will claim an impressionist painting, like say a Gauguin, is an accurate representation of what the artist saw with his own eyes:



But is it art and even great art, art I appreciate? For sure.

From experience and general feel, what I show should sound pretty similar to "real tubes", specifically something like this:

https://www.tangible-technology.com/tubes/1566/1566.html

The JFET circuit I show will clip earlier than the tube circuit. I see that as problem. I like more voltage on gain stages, when SPL bang on about "high voltage" circuits they are not just Tub Thumping and Welkin Ringing, but they are onto something. The same discrete Op-Amp at +/-15V and +/-60V sounds surprisingly different and the high voltage operation sounds more tube like, in the positive sense of the word.

If the circuit was adapted to run on 48V it should probably clip way before the ADC and it would be ahead of the 24V one.

An output transformer could (SHOULD?) be added. Elaborations can be made to make the follower more transparent.

I'm ok to help a little, once you can really define clearly what result you want.

Thor

 

[Emmathom]

Hi @thor.zmt
I happily accept a helping hand in the creation of this jfet preamp. Thank you !
What result am I looking for? Well... I can summarize certain points:

- +20~+30dB from the circuit knowing that the OEP X187B input transformer already offers +20dB

- no symetrical output (so no output Tr) because +50dB is enough gain and my sound card is in the same flight-case as the pre's (short distance) and I don't want to buy 2 good output transformers (expensive to get good quality)

- a "not too complicated" circuit (without a wheelbarrow of components) so in the spirit of what you have already posted

- for the supply voltage, a 2*18v • 330mA transformer is already installed (for the BJT circuit) as well as a 12v • 1300mA (300mA are already consumed by the heating) and finally a Meanwell RS-15 / 48v smps • 300mA for phantom is also present in the device.

- in terms of "sound": no need for exaggerated saturation, good SNR, personnality (?)

The jefts already in my stock :
- K170 (4 with same IDSS & VGS)
- K30A (10)
- J113 (30)
- J305 (2)
- J310 (3)
- 2N5486 (8)
- MPF102 (10)
Mosfets N in stock : IRF840 / IRF 820 / IRF620

*Gauguin is a wonderful artist

Best Regards

 

[Emmathom]

some process 55...

 

[k brown]

Well, IT DEPENDS.

You might want a very colourful Mic Pre. Then Hampton, with "sonically colourful" passive parts like Allen Bradly or Xicon Carbon composition resistors, Mylar or polycarbonate capacitors and coloured electrolytic types like Axial Nichicon VX might be the way to go. Add very "Iron sound" input transformers (tiny) and a small size low nickel core output transformer.

As said, sonically it would be a caricature of a good tube circuit, everything overly emphasised, kinda like KFC "secret spice" (a truckload of MSG and deep/pressure fried in industrial oil) vs good southern fried chicken without MSG, Trans fats etc. et al....



I am not saying that a "colour preamp" is not valid, it is valid. Recording is Art, not science. Nobody will claim an impressionist painting, like say a Gauguin, is an accurate representation of what the artist saw with his own eyes:

View attachment 126349

But is it art and even great art, art I appreciate? For sure.

From experience and general feel, what I show should sound pretty similar to "real tubes", specifically something like this:

https://www.tangible-technology.com/tubes/1566/1566.html

The JFET circuit I show will clip earlier than the tube circuit. I see that as problem. I like more voltage on gain stages, when SPL bang on about "high voltage" circuits they are not just Tub Thumping and Welkin Ringing, but they are onto something. The same discrete Op-Amp at +/-15V and +/-60V sounds surprisingly different and the high voltage operation sounds more tube like, in the positive sense of the word.

If the circuit was adapted to run on 48V it should probably clip way before the ADC and it would be ahead of the 24V one.

An output transformer could (SHOULD?) be added. Elaborations can be made to make the follower more transparent.

I'm ok to help a little, once you can really define clearly what result you want.

Thor

Well, just to be 'that guy', Gauguin was technically Post-Impressionism; often referred to as Symbolism or Synthetism.

 

[thor.zmt]

- +20~+30dB from the circuit knowing that the OEP X187B input transformer already offers +20dB

- no symetrical output (so no output Tr) because +50dB is enough gain and my sound card is in the same flight-case as the pre's (short distance) and I don't want to buy 2 good output transformers (expensive to get good quality)

I would suggest two stages with 20dB each, first stage directly driving the Gain control without buffer and second stage with buffer to drive a line out.

We can make the first stage switchable gain, 0/10/20dB.

For improved utility I would suggest to use a similar scheme as the Altec Pre Eddie C. mentioned, so here:


Instrument in (using a separate toggle), no output transformer.

- a "not too complicated" circuit (without a wheelbarrow of components) so in the spirit of what you have already posted

Let's go with that.

- for the supply voltage, a 2*18v • 330mA transformer is already installed (for the BJT circuit) as well as a 12v • 1300mA (300mA are already consumed by the heating) and finally a Meanwell RS-15 / 48v smps • 300mA for phantom is also present in the device.

Let's use 48V, it would make it a project others can replicate, perhaps with higher gain and output transformer.

- in terms of "sound": no need for exaggerated saturation, good SNR, personnality (?)

So not a Picasso or Kandinsky then....

The jefts already in my stock :
- J113 (30)

Let's use those.

Mosfets N in stock : IRF840 / IRF 820 / IRF620

Let me see which of these has the least Crss, alternatively, if you have (say) a few BD140 around we can use J113 and BD140 top make a really great buffer stage.

So:


MIC (with Phantom power) -> TX 1:10 -> Pad 0/-10dB -> Gain stage 10dB/20dB -> Gain Control 10K(A) -> Gain Stage 20dB fixed -> Buffer -> Output


Sounds Good?

Thor

 

[Emmathom]

Sounds very good Much thanx !
I have BD140 in stock so J113.
Ok for using +48v

Some remarks :
- as Borbely propose, maybe 2 jfets in // at 1rst level to reduce capacitance ?
- Tr In will be at the rear / Pot & all switches at the front : so with a pad switch at the front panel of the device, isn't any trouble to make the signal go from back to front then front to back ? about 50cm full length... (can use shielded cable if needed anyway)
- buffer with a cascade Jfets ? as Pass do ? (see attached) that's what is set in my tube preamps... > I think you mean BD140 as buffer...

 

[Emmathom]

so as not to be "passive" and wait for you to do everything, I drew a sketch of what the outline could be...
not sure we need a pad and I will not install the Hi Z instrument input (because I've several good DI units)

 

[Voyager10]

Our replies have crossed...
So better try this schematic...

Just as a suggestion (I haven't simulated it) with this circuit.

The gain is set (mostly) by the ratio of the 1k resistor R2 to R8 / R9 effectively in parallel (i.e. 28K). We could replace R8/R9 with a single resistor + DC blocking capacitor to ground, then make that single resistor switchable (say 2k7, 10K, 27k) to switch the gain in roughly 10dB steps.

The point of this would be to reduce the voltage swing on the FET drain for large signals, which might (um, simulation needed) give lower distortion compared to +26dB gain then attenuation with a volume pot.

You could do the same trick with two gain stages (say a 2 pole 6 way switch) to have a very wide switchable gain range, Neve 1073-style.

 

[Emmathom]

Just as a suggestion (I haven't simulated it) with this circuit.

The gain is set (mostly) by the ratio of the 1k resistor R2 to R8 / R9 effectively in parallel (i.e. 28K). We could replace R8/R9 with a single resistor + DC blocking capacitor to ground, then make that single resistor switchable (say 2k7, 10K, 27k) to switch the gain in roughly 10dB steps.

The point of this would be to reduce the voltage swing on the FET drain for large signals, which might (um, simulation needed) give lower distortion compared to +26dB gain then attenuation with a volume pot.

You could do the same trick with two gain stages (say a 2 pole 6 way switch) to have a very wide switchable gain range, Neve 1073-style.

Hi
which diagram do you refer to ? this one below from Thor ?

 

[rankot]

@rock soderstrom : since the 2N5457 seems to be out-of-stock at Mouser's could you please give a equivalent ref. Jfet ? (I've got MPS102 & J113 in stock)

It is obsolete. The best way to obtain is to buy SMD version, MMBF5457. They are also better in means of consistency for some reason.

 

[Voyager10]

Hi
which diagram do you refer to ? this one below from Thor ?

Yes, that's the one, if that's still of interest.

 

[thor.zmt]

The gain is set (mostly) by the ratio of the 1k resistor R2 to R8 / R9 effectively in parallel (i.e. 28K). We could replace R8/R9 with a single resistor + DC blocking capacitor to ground, then make that single resistor switchable (say 2k7, 10K, 27k) to switch the gain in roughly 10dB steps.

The point of this would be to reduce the voltage swing on the FET drain for large signals, which might (um, simulation needed) give lower distortion compared to +26dB gain then attenuation with a volume pot.

That does not work. The distortion is caused by the curvature of the JFET's transconductance curve. So it is percentage of input swing. The correct way is increased degeneration.

You could do the same trick with two gain stages (say a 2 pole 6 way switch) to have a very wide switchable gain range, Neve 1073-style.

Well, I am suggesting something like this. But in my case increased degeneration, which also sadly increases noise.

All a tradeoff.

Thor

 

[thor.zmt]

so as not to be "passive" and wait for you to do everything, I drew a sketch of what the outline could be...
not sure we need a pad and I will not install the Hi Z instrument input (because I've several good DI units)

I will show them, you can always omit them.

Wait a bit.

Thor

 

[Voyager10]

Oh, I was thinking we wanted to keep (some) distortion from the quadratic Vgs-Id curve, but other sources of distortion (e.g. from the voltage follower, capacitance changes with Vds, or good old clipping) are best miniised.

If not, fair enough, it was just a suggestion.

 

[thor.zmt]

Okay, here my suggestion:



Some comments.

The LF Rolloff is mainly the input Transformer's limited inductance.

Maximum gain for the complete Mic-Pre is 60dB.

The buffer circuit uses what is variously called JFET/BJT Sziklai, Inverted Darlington etc. It runs at ~20mA and the 1K tail resistor should be at least 1W.

The output gain stage has more gain than input, this relates to relative overload levels, determined by the ~1V Bias of our J-Fet's. The J113 have less bias than J112 or J111. So we need to keep the signal at the gate at around 0.5V RMS or less to keep in the mostly linear range. The greater the input signal, the greater the distortion. Once the gate is driven positive relative to the source the input signal is clipped hard by the Gate diode, this happens at a bit over 1V RMS.

The actual gain circuit is a modification of a Tube circuit called "Mu Follower". As JFET's are more Pentode than triode like, the AC load at drain of the lower JFET determines the gain together with the source resistor, which set's the JFET's effective transconductance. The upper JFET acts as an AC constant current source or as it is sometimes called a "virtual choke"

The resistors R9/10 set the gain together with R2 for the output stage. Gain is around 32dB. Maximum undistorted output into a 10k load is around +21dBm or 8V and for +20dBm we see around 0.1% THD.

As the input stage has limited input range and the intended transformer has a relatively high step-up ratio (9.13) a signal of -14dBm (~ 155mV) will overload the input stage. For a "bogey" Microphone with 10mV/Pa sensitivity this only allows ~118dB SPL before overloading the input stage. With the 10dB and 20dB pad higher sensitivity Microphone's and high SPL's are covered.

One channel will need 48V at appx. 40mA (including Phantom Power). Before anyone tells me that I have the wrong resistors for Phantom power, I do not have. The 1k filter resistor in the power supply line must be accounted for. This means we want 4.8k Resistors and we use 4.7k as NPV.

The 1mH / 10,000uF Power supply filter has a -3dB point at around 2Hz (that is the 9 Ohm DCR of the choke) and is 35dB down at 100Hz. With all the added RC filtering we have around 90dB PSRR on the raw 48V line at 100Hz, enough to eat up any reasonable noise from the PSU.

That's it. I hope that six active devices for a total of ~42dB active gain with relatively low distortion, good output levels and drive is simple enough.

I think a 2:1 Output Transformer (e.g. OEP Z30A11C) would be a neat option for a true stand alone unit. This would limit available output to +15dBm (with an ideal Transformer) . To compensate the 6dB lost gain from the output transformer, the Gain control can be changed to 20kOhm (even 50k, giving up to 68dB gain). This could look something like this:



68dB max gain, maximum output +18dBu @ 1% THD (not accounting transformers), with gain at 48dB and 0dBm out, SNR is around 81dB, this would need 3mV input (or -48dBm) which would with a "bogey" Mic happen at ~ 84dB SPL. Mild tube like harmonic distortion, comparable to that of classic open loop tube circuitry.

Will it sound any good? It would need building and listening. I'm willing to bet a bottle of Thai Chang Beer it will sound good.

Thor

 

[thor.zmt]

Oh, I was thinking we wanted to keep (some) distortion from the quadratic Vgs-Id curve

There is so much, compared to tubes, that we need to work hard to keep everything under control.

You can see a fully elaborated and more optimised version above.

Thor

 

[Emmathom]

Thank you +++ Thor for this great work !
I'm far from being as experimented as you are (!!!) so I'll take some time yo study your schematic patiently... and make my list of components I'm missing.
(btw : your link "126444" doesn't work)

At first look, some remarks & questions :

- this circuit is gonna share 48v with the 2 tube channels already installed and my phantom power is far more "basic" than yours : PSU (smps) out > 100Ω > switch with 100uF to ground > 6,8KΩ to each XLR Pin 2 & 3 : will this be a problem ?

- the setup of my X187B is also basic as there are no 10pF In/Out / no R serie / no inductance at all (never seen this arrangement)

- 50v capacitors seem just enough to me : I would put 63v... ?

- for what I understand, the input charge seen by the primary of the X187B is (R20 + R28 + R5)/83 (9,13 square) > 48,1KΩ/83 = 580Ω mic charge ? or the charge is made by L1 ? don't understand this part...

- what is Z output of this circuit ?

Best Regards