Jfet pre

[Emmathom]

I forgot something : my 30 J113 have quite closed IDSS & VgsOff but I supposed I don't have to match specifications for this circuit ?

 

[thor.zmt]

(btw : your link "126444" doesn't work)

????

- 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 ?

Not specifically a problem. I would recommend more filtering like I show, with 1k you get 20dB more noise filtering.

- 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)

They are the parasitic components INSIDE the transformer. In a simulation they must be accounted for. I find it easier to do that with extra components then try to make the transformer models required otherwise.

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

Nobody stops you. But the capacitors are in a very low stress application, no appreciable heating from currents. So they will be fine I think.

Nichicon KZ come in 25V/50V/100V, not 63V. They, like the ES are part of the Muse range.

- 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...

The actual impedance is in the datasheet:

https://carnhill.co.uk/wp-content/uploads/2019/07/X187B.pdf

It is the 47k load line.

- what is Z output of this circuit ?

Same as the build-out resistor, so around 51 Ohm.

Thor

 

[thor.zmt]

I forgot something : my 30 J113 have quite closed IDSS & VgsOff but I supposed I don't have to match specifications for this circuit ?

Matching should not be needed, but it will do no harm to have the lower JFET's matched between channels.

Thor

 

[Emmathom]

Ok !
I've got in stock :
- BD140 (10 items)
- J113 (30 items)
- 1uF 50v Wima MKS2 (10 items)
- 1uF axial PIO russian (2 items)
- 1 uF "can style" old PIO russian (2 items)
- 22uF 16v Nichicon Muse BP (10 items)
- most of the resistors (will put 2x2,2KΩ 1w in // for BD140 collector if it's ok)

* C12 is 10.000uF 50v ? it's written 10m...

I must order some R (3,3MΩ / 3,6MΩ)
I will build this circuit on a "pre-drilled board" with ptp wiring with silvered-cooper PTFE 26AWG > I like to work so

 

[Emmathom]

I want to sincerly thank you Thor for your help...
It's 9h30 pm now in France and I got to leave the computer, but tomorrow I'll be back !

 

[thor.zmt]

- 1uF 50v Wima MKS2 (10 items)
- 1uF axial PIO russian (2 items)
- 1 uF "can style" old PIO russian (2 items)

Yuck, on all. Please get some MKP02 or similar.

- 22uF 16v Nichicon Muse BP (10 items)

Not enough voltage. Use 50V.

- most of the resistors (will put 2x2,2KΩ 1w in // for BD140 collector if it's ok)

Yes, fine.

* C12 is 10.000uF 50v ? it's written 10m...

Quirk of the software. 10,000uF = 10mF. The latter is correct according to SI. But EE's are quirky.

I will build this circuit on a "pre-drilled board" with ptp wiring with silvered-cooper PTFE 26AWG > I like to work so

I often do the same. I have reels to this wirewrapping wire in the shop.

Thor

 

[Emmathom]

Shopping cart at Banzai (https://www.banzaimusic.com/home.php)
no MKP2 1uF and MKP10 will be too much so I found Jb MKP "according to my budget" (or can it be Audyn Cap Q4)

 

[thor.zmt]

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).

...

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.

Ok, I looked a a bit more on small details. I tend to use stepped controls for gain, so the most likely (low cost) would be an 11-position Lorlin switch for gain. Essentially this:


They are not as cheap as they used to be, but they are still cheap, cheerful and will do the job. Of course with only 10 attenuation positions and 3dB steps, we get only 30dB maximum attenuation.

So a wider range coarse volume is probably desirable. I added a combination of Pad's before and after the first active stage, allowing 0dB to -40dB using a 2-Pole 5-Position Lorlin switch, so in combination we have 70dB range of gain adjustment in 3dB steps. Wiring of the switches should be obvious, but ask if not.

Of course, you could instead use the ELMA 47 Position switched attenuators, which allows 70dB in 1.5dB steps, but it's kinda dear. Here the idea is to be cheap, hence OEP Transformers, JFET's instead of tubes and minimal boutique parts.

In the end we get to this:



Most parts should be available from Mouser. J113 shows availability, as does most of the rest. OEP Transformers are widely distributed internationally. Veroboard should be fine to build this, no need for a dedicated PCB.

Thor

 

[thor.zmt]

Shopping cart at Banzai (https://www.banzaimusic.com/home.php)
no MKP2 1uF and MKP10 will be too much so I found Jb MKP "according to my budget" (or can it be Audyn Cap Q4)

Audyn seems overkill. But can use.

How about Mouser?

https://www.mouser.com/ProductDetail/WIMA/MKP2C041001N00MSC9?qs=RB4whv9F6rxrGnrhoRYQ/w==

Thor

 

[Emmathom]

Last post before bed :
* Mouser's shipping to France is 20€ / Banzai is 9€ and I can find all I miss there so...
* Could you tell me please what is total current consomption per channel ? because I guess I should install a RC filter to "isolate" each channel from SMPS output (and from phantom power too), for exemple : 5Ω + 10.000uF per channel (R value will depend on mA drawn per each channel) > (5Ω offers -3dB at 3Hz with 10.000uF = -0,5v if 100mA per channel)

I will come back to you about the pads tomorrow (some requests)
Good evening

 

[Emmathom]

20Hz response is not the ultimate factor there.
Electrolytic capacitors introduce distortion when submitted to a voltage.
In order to minimize it, it is necessary to use a value much higher than the one that gives a good LF response.
The old rule of thumb that said "10 times the value needed for a specific -3dB response" - which, BTW, is coincident with -0.5dB - is no longer adequate for the level of distortion expected for current designs.
I commonly use values that are about 5-10 times that.

Hi @abbey road d enfer, we use a smps PSU so no 100Hz to filter here. And the current consumption is quiet low (would say not more than 100mA / 48v for 2 channels so...)
The "theory" is 160.000uF per Ω serie for -3dB at 1Hz (without consedering high current consumption)
Thor put 10.000uF just after the PSU so it only needs 16Ω for a good RC filter.
After that, 2,2kΩ + 100uF per stage is more than needed (160mF/2200Ω) = 72uF

If I follow the "10-times-rule" with an HT PSU feeding anodes that need 10mA under 300v (for ex.) : we don't want to rise in impedance, so we fix 320Ω > (160mF/320Ω)*10 = 5000uF !
Never seen such value in any tube preamp PSU...

In my tube pre's I aim -3dB at 3Hz and so according to our exemple > ((160mF/3)/320Ω) = 54mF/320Ω = 170uF and I put (more-or-less) 3 times to have a good energy stocked (so 500uF) and never got any "hum". The capacitance value increases 3 times from R to R going from PSU to circuit, like (for ex.) 47uF > 150Ω > 120uF > 150Ω > 330uF. The first cap must recharge quickly because of audio signal current draws (more in amplifier than in preamp) while the followers stock gradually more current.
That's my p.o.v. but I'm maybe doing wrong (?)
BR

Attached GYRAF G9 PSU

 

[Emmathom]

Hi !
@thor.zmt : I dont intend to criticize your work, but why all these pads ? do we really need all of them ? it complicates the circuit...
We could have a -20dB switch pad at input as you previously drew (15KΩ in serie with 33KΩ fixed to ground) in case of the source is too high and then keep the 10KΩ log pot (I have some in stock) and that's all folks.
I search simplicity. The less components the better it sounds (I think so).

I don't intend to buy output Tr (and I don't have space to install a pair in the box)
Your first schematic fits my goal I guess.

But I intend to follow your directions as best as possible as you are the "father" of this realization

I come back on the Tr In secondary "R impedance set" : the Tr is a 1:9 and with 33KΩ at secondary, the primary "sees only" 33KΩ/81 = 400Ω (?) or I miss something...
I generally put 220KΩ ~ 300KΩ on my tube pre's for the mics to see 10 times their own impedance (generally 200Ω)

Best Regards

 

[Voyager10]

The load across the transformer secondary is 33K + 10K + 5.1K in series, or 48.1K. This matches the X187B's intended secondary load impedance (nominally 50K).

Maybe a simpler option for pads is one DPDT switch to take the first gain stage out of circuit and connect the transformer secondary straight to the gain pot (better make it 50K again). The maximum input level is limited only by the transformer then.

One minor point: the phantom power switch could be the other side of the 1K resistor. As shown it will short out C9 (100u charged to 48V) if you turn it off when the circuit is powered.

 

[Emmathom]

The load across the transformer secondary is 33K + 10K + 5.1K in series, or 48.1K. This matches the X187B's intended secondary load impedance (nominally 50K).

Ok if the datasheet says so ! But I have put 220KΩ on the 2 tube channels (?) according to the calculations I made (primay goal 2KΩ x Tr ratio square > 2000x81 = 160KΩ (at least)
So what is pros/cons of such a 220KΩ value ? Should I mofify ?

Maybe a simpler option for pads is one DPDT switch to take the first gain stage out of circuit and connect the transformer secondary straight to the gain pot (better make it 50K again). The maximum input level is limited only by the transformer then.

A 50KΩ pot could introduce noise : I prefer a -20dB pad by switch at input + 10KΩ pot interstage if you don't mind

One minor point: the phantom power switch could be the other side of the 1K resistor. As shown it will short out C9 (100u charged to 48V) if you turn it off when the circuit is powered.

As I mentionned earlier my phantom power is already installed in this quad pre : smps 48v > 100Ω each switch + 100uF 63v for filtering > 2x6,8KΩ to XLR pin 2 & 3

Actual wiring below

 

[Voyager10]

It's entirely your build!

If you've used the same transformer with a 200K+ secondary load before, and it works OK, there should be no problem using a FET instead of a tube. Likewise, if you have phantom power which works, there's no need to change it.

(For a 220K secondary load, using 150K, 47K and 22K instead of 33/10/5.1 for the pad will do the job).

 

[thor.zmt]

we use a smps PSU so no 100Hz to filter here.

Don't bet on it. There tends to be a fair bit.

And the current consumption is quiet low (would say not more than 100mA / 48v for 2 channels so...)

I stated it in the text, 48V/40mA Channel including Phantom Power (assuming a pull-down to 12V by the Mic).

The "theory" is 160.000uF per Ω serie for -3dB at 1Hz (without consedering high current consumption)
Thor put 10.000uF just after the PSU so it only needs 16Ω for a good RC filter.

I specified an Inductor, Abracon which is 9 Ohm DCR, 1mH. Together with a 10,000uF/10mOhm ESR Electrolytic cap it should kill a lot of noise.



As visible, at the typical 67kHz switching frequency we get > 90dB attenuation, at 100Hz we 35dB attenuation. The ripple at 67kHz at rated current is often as high as 100mV P-P while the 100Hz component can often be as high as 10mV P-P.

After that, 2,2kΩ + 100uF per stage is more than needed (160mF/2200Ω) = 72uF

No, it is fine. Here the cascaded Filtering:

If I follow the "10-times-rule" with an HT PSU feeding anodes that need 10mA under 300v (for ex.) : we don't want to rise in impedance, so we fix 320Ω > (160mF/320Ω)*10 = 5000uF !
Never seen such value in any tube preamp PSU...

The current in the first stage is 0.5mA and in the second stage 2.5mA including ~0.35mA for the Loading resistors.

Also, due to the nature of the active circuit (Virtual inductor) it is not very sensitive to PSU Impedance.

I dont intend to criticize your work, but why all these pads ? do we really need all of them ? it complicates the circuit...

I stated that these are options.

A lot depends on context and use. I personally like good control over things like overload margins, noise and gain.

We could have a -20dB switch pad at input as you previously drew (15KΩ in serie with 33KΩ fixed to ground) in case of the source is too high and then keep the 10KΩ log pot (I have some in stock) and that's all folks.

All up to you.

I search simplicity. The less components the better it sounds (I think so).

Up to a point - yes.

I gave you a simple circuit meant for your specific state requirements in post # 58, this:



You can use the -20dB Pad only.

I then proceeded to extend the design with Output Transformer and more specifics on how I PERSONALLY would implement this, in case someone else is interested to make a more flexible and slightly more complex stand alone version, that drops in for any old mic pre and dos not break the bank with expensive components.

Hence polarity, a 30dB range stepped attenuator in 3dB for fine gain and a 40dB Attenuation in 10dB steps.

It should be possible to balance things like intentional input overload vs output well (this assumes an ADC with ~+20dBm for 0dBFS).

I come back on the Tr In secondary "R impedance set" : the Tr is a 1:9 and with 33KΩ at secondary, the primary "sees only" 33KΩ/81 = 400Ω (?) or I miss something...

Well, I use 47k.

I generally put 220KΩ ~ 300KΩ on my tube pre's for the mics to see 10 times their own impedance (generally 200Ω)

Using the OEP transformer you specify this likely leads to HF peaks. It is designed and specified for 12k....47k load. The very limited primary inductance and seemingly high parasitic capacitance means that even with a 47k load we approach 600 Ohm only in the midrange, at 20Hz and 15kHz it has dropped to 500 Ohm.

Maybe a simpler option for pads is one DPDT switch to take the first gain stage out of circuit and connect the transformer secondary straight to the gain pot (better make it 50K again). The maximum input level is limited only by the transformer then.

Yes, this is an option. I'm a little pernickety about polarity, if doing this the transformer secondary should be inverted.

One minor point: the phantom power switch could be the other side of the 1K resistor. As shown it will short out C9 (100u charged to 48V) if you turn it off when the circuit is powered.

Yes, it was my intention to have a hard short, it makes not much difference.

Here something for the "big" design that incorporates your suggestions. I added one more Gain step as jut Lo/Hi feels a bit of a big gap. A Lorlin Rotary could be used, or two toggles.

A 50KΩ pot could introduce noise : I prefer a -20dB pad by switch at input + 10KΩ pot interstage if you don't mind

I ran this, for 0dBm (.775V) out with Output Transformer without transformer) the simulator suggests 85dB SNR, which is fine.

If the output transformer is omitted and the gain should be at 60dB max. then change as follows:



Thor

 

[thor.zmt]

Actual wiring below

Using the center-tap is not such a great idea, UNLESS the primay winding is completely symmetrical.

The datasheet does not state resistor match across the two halves of the primary.

Also, if using centre-tap, the resistor values are 3.3k + 100R.

220k on the secondary MAY extend the high frequencies or it may lead to peaks.

A final "Emmathom" version:



Gain is 60dB max. with -14dBm maximum input levels, the 20dB pad gets us 40dB max gain and +4dBm maximum input levels. Load on transformer is 178k, the closest to 220k that gives -20dB with standard values.

Thor

 

[Emmathom]

Thank you @thor.zmt for this great job ! I like the last version and will stick to it.

"Gain is 60dB max. with -14dBm maximum input levels, the 20dB pad gets us 40dB max gain and +4dBm maximum input levels. Load on transformer is 178k, the closest to 220k that gives -20dB with standard values" > that suits me well !
I just have to drill a 6mm hole for a switch on the front pannel (no big deal)

And I will change my 3,9KΩ by 3,3KΩ (there is 100Ω is before every switch - the actual configuration is : smps 48v > 100Ω > switch + 100uF • 63v > 3,9KΩ > Tr center tap). The other solution is to put 6,8KΩ on Pin2 & 3 and forget the center tap (?)

Considering what I just learned about input transformers, may I ask you one more advice please ? in my quad pre's I use CAPI CA2622 with (again !) 220KΩ secondary charge...
- does it cause any problem ?
- what value should I install ?
- what would be the benefit ?

In my tube circuits (12AY7 + 12AU7 AF) I didn't install any gain level (I think I'm gonna change this) but a 100KΩ pot. Log (with 150pF from in to out) after the 12AU7 and then a cascade jfets Z adaptation. It sounds very well.

BR

 

[rock soderstrom]

A final "Emmathom" version:

Very nice. Unlike other gods, the almighty Thor is not stingy with concrete recipes here. I would like to thank him for that and also for the Tina tip! I think I'll etch a PCB and see what the circuit can do for me. I'm curious how the "overdrive" behavior is, that's where the HamptoneFET shines. Thanks again.

 

[Emmathom]

Sorry @thor.zmt bit is this schematic for your tests or do I install it as it is ? [10Ω > 1mH] + [10mF + 10MΩ to ground] ?
Because it is not as so on the final schematic...