Oktava 012: building a new preamplifier

[Gus]

Why a 170?  did you not read the thread?  Rossi and ricardo have taken the time to post some good information.
 
I stayed out of this thread till now maybe you should do a good web search and a good search of this forum for things about the 012, JFETs and capacitors.  The tricky part is figuring out what is good information

 

[mark88]

Why a ls170?  did you not read the thread?  Rossi and ricardo have taken the time to post some good information.
 
I stayed out of this thread till now maybe you should do a good web search and a good search of this forum for things about the 012, JFETs and capacitors.  The tricky part is figuring out what is good information

Rossi suggested not to move from the original schematics.
The Toshiba 2SK170 BL is the element specified by Scott Dorsey, used in the deluxe modification by Bill Sitler, and used in the new OKtavas...
Which transistor do you suggest?

 

[Rossi]

NPO/COG ceramics do what capacitors are supposed to do AND NOTHING ELSE.  They don't sound brighter, darker bla bla.

This stuff about Silver Mica et al is tested .. including the propensity of Polystyrenes to microphony.  You can test this yourself.  Replace your capsule with a (shielded) polystyrene capacitor of the same capacitance and tap the resulting assembly.  You need to use a circuit that doesn't have other microphonic bits in the signal chain.

Guru Scott Wurcer did this for his Linear Audio articles but the results appeared in MicBuilders.

I did some tests this morning: NP0 is indeed the least microphonic and polystyrene was the most microphonic. Polyester and polypropylene were slightly more microphonic than NP0 ceramic but not much. However, none of the capacitors tested was problematic. Yes the meters move a bit when you tap the polystyrene one, but it I found it impossible to excite it with sound waves (I'm glad nobody was around to see me screaming at at caps on a breadboard.  : ) But lowest microphony is not the ultimate goal; sound is.

I couldn't find the Wurcer Scott stuff on micbuilders, but I know the Bateman articles, of course. I'm averse to esoteric thinking, and I'm one of the few guys here who does electric and acoustic microphone measurements. So I'm not in the snake oil business, but my own findings only partially conform to what I've read on cap distortion etc. There are areas where I prefer to rely on my ears, and one of them is caps.

If you play guitar, try this: exchange the tone pot cap for a polypropylene one (I tried an WIMA FKP 1.5 nF in a Strat, which should have produced a more P90-ish sound with the tone control on 1)). Sounds terrible even with the tone control on 10. I have no explanation for this.

 

[Rossi]

Rossi suggested not to move from the original schematics.
The Toshiba 2SK170 BL is the element specified by Scott Dorsey, used in the deluxe modification by Bill Sitler, and used in the new OKtavas...
Which transistor do you suggest?

As I've said before, read up on the 2sk170, we've had lengthy discussions around here.

Originally Oktava used some Russian FET. If they switched to the 2sk170 it was because of the Dorsey recommendation. The article is quite old by now, I doubt Dorsey would still recommend it. That's the curse of a journalist (I am one myself): You live and learn, but what's written is written.

The J305 Ricardo suggested is a much better choice for this circuit.

 

[mark88]

Rossi suggested not to move from the original schematics.
The Toshiba 2SK170 BL is the element specified by Scott Dorsey, used in the deluxe modification by Bill Sitler, and used in the new OKtavas...
Which transistor do you suggest?

As I've said before, read up on the 2sk170, we've had lengthy discussions around here.

Originally Oktava used some Russian FET. If they switched to the 2sk170 it was because of the Dorsey recommendation. The article is quite old by now, I doubt Dorsey would still recommend it. That's the curse of a journalist (I am one myself): You live and learn, but what's written is written.

The J305 Ricardo suggested is a much better choice for this circuit.

thank you. but, as suggested, I've read a half dozen post on this forum regarding the 2SK170 and there are not a clear recommendation to substitute it with another fet. There's also a post which says that moving to a faster FET would increase the response and the detail, but this could modify the creamer sound of the Oktava into a brighter one... which I'm not looking for.
There's also to say that there are a normal 2SK170 version, and the 2SK170BL which has different properties (faster than the non-BL version).

It is also clear that I have no experience here so I trust completely on you recommendations!
Thanks for your patience

 

[mark88]

Another question: I read an interesting post. Could I improve my project with this design?

Re: DIY home made P2P Oktava Mk012.......circuit questions & observations
« Reply #55 on: May 28, 2011, 19:01:22 »

    Like
    Quote

Hey Gus! Thanks for all the info.

I did handle the input cap and input resistors with my fingers. I did clean connections and leads with a flux remover pen/isopropyl, but should I also use a little isopropyl to clean the casings? My P2P version is very quiet noise wise, and the output is still quite low. I used a Xicon polystyrene as the input cap. I'll also try cleaning the capsule connections.

Couple of Q's. I know the 1G/input cap/capsule/FET gate connections need to be made in the air, but is it essential that the input cap connections be soldered on the bottom? I ask because using a polystyrene, I need to mount it on top, and then run the leads underneath, which makes the leads quite long. I have some single end/double turret push pins that I could use if it's ok to just connect it all on the top.

The phantom power. I measure 40VDC at pin 3, but only 10VDC on the capsule side of the 30M resistor. I did experiment with replacing the 30M with a 10M which gave me 24VDC, and it did increase the output marginally. I did not check to see if this impacted other voltages on the board (doh!). The Apex 180 schematic shows a DC to DC converter, and a 1M resistor prior to the 1G resistor and capsule. I was going to rebuild one of them to get some voltage measurements to compare.

tchgtr- Thanks man. I have to look through my inventory and see what I got. I think I have a .033 cap to try. That sounds like a good starting point.

 

[ricardo]

I did some tests this morning: NP0 is indeed the least microphonic and polystyrene was the most microphonic. Polyester and polypropylene were slightly more microphonic than NP0 ceramic but not much. However, none of the capacitors tested was problematic. Yes the meters move a bit when you tap the polystyrene one, but it I found it impossible to excite it with sound waves (I'm glad nobody was around to see me screaming at at caps on a breadboard.  : ) But lowest microphony is not the ultimate goal; sound is.

I couldn't find the Wurcer Scott stuff on micbuilders, but I know the Bateman articles, of course. I'm averse to esoteric thinking, and I'm one of the few guys here who does electric and acoustic microphone measurements. So I'm not in the snake oil business, but my own findings only partially conform to what I've read on cap distortion etc. There are areas where I prefer to rely on my ears, and one of them is caps.

Why don't you listen to your microphonic caps?

Use headphones or record a bit when you tap the mike or shout at it.  As I said, you need to make sure the rest of your circuit has no microphonic bits.

I first came across this with polystyrenes circa 1980 when doing the circuit for the Mk4 Calrec Soundfield.  The experimental circuit was called the Calrec Seismograph. 

Microphony with Polystyrenes is VERY high Q so sounds 'metallic'.  You can work out how far below the 'acoustic' signal it appears ... and decide whether you will hear it or like it.

Wurcer saw it cos he was doing noise spectrums and saw these huge spikes rising up from the noise floor.  Polystyrene microphony is due to the construction so some are better than others but all are worse than NPO/COG ceramics.

On a separate note, you should listen to the noise spectrum of important electrolytics in the signal chain of a very LN amplifier eg a ribbon mike preamp.  Doing this led to my prejudice against Tantalums.  They have a crackly spiky noise .. a bit like insufficient dither in a digital chain.  Maybe that's the old Neve sound 

There's also to say that there are a normal 2SK170 version, and the 2SK170BL which has different properties (faster than the non-BL version).]There's also to say that there are a normal 2SK170 version, and the 2SK170BL which has different properties (faster than the non-BL version).

If you insist on using 2sk170, use 2sk170VG.

These are hand carved from solid Unobtainium by Virgins.  Apart from being even faster than J305, they have increased Clarity & Definition though slightly noisier than J305, a small price to pay.

 

[Rossi]

Well, I don't like polystyrene anyway, so no immediate need to test it any further. However, polystyrene was used in many very respectable vintage mics. These days I see polyprop in most higher end mics.

Ceramic caps can become microphonic, too. In older U47 mics a ceramic cap (I don't know which type) was used as a filter cap for the polarization voltage. 50 years later, most of them have become microphonic. The later used polyester caps are usually fine even today.

 

[ricardo]

However, polystyrene was used in many very respectable vintage mics.

The reason was simply 1n NPO/COG ceramics were Unobtainium in 1980

Ceramic caps can become microphonic, too. In older U47 mics a ceramic cap (I don't know which type) was used as a filter cap for the polarization voltage. 50 years later, most of them have become microphonic. The later used polyester caps are usually fine even today.

In most mike circuits, the decoupling caps ARE in the signal path.  Using MLC ceramics for decoupling without nice electrolytics is asking for trouble.

In most cases, nice electrolytics WITHOUT ceramics or Golden Pinnae stuff in parallel work well.

The last cap decoupling the polarizing voltage to the capsule is ALWAYS DIRECTLY in the signal path.  It needs to be about 100n so a NPO/COG ceramic would be too big.  Polyesters and even some Golden Pinnae stuff are good for these.

 

[Gus]

Rossi suggested not to move from the original schematics.
The Toshiba 2SK170 BL is the element specified by Scott Dorsey, used in the deluxe modification by Bill Sitler, and used in the new OKtavas...
Which transistor do you suggest?

As I've said before, read up on the 2sk170, we've had lengthy discussions around here.

Originally Oktava used some Russian FET. If they switched to the 2sk170 it was because of the Dorsey recommendation. The article is quite old by now, I doubt Dorsey would still recommend it. That's the curse of a journalist (I am one myself): You live and learn, but what's written is written.

The J305 Ricardo suggested is a much better choice for this circuit.

thank you. but, as suggested, I've read a half dozen post on this forum regarding the 2SK170 and there are not a clear recommendation to substitute it with another fet. There's also a post which says that moving to a faster FET would increase the response and the detail, but this could modify the creamer sound of the Oktava into a brighter one... which I'm not looking for.
There's also to say that there are a normal 2SK170 version, and the 2SK170BL which has different properties (faster than the non-BL version).

It is also clear that I have no experience here so I trust completely on you recommendations!
Thanks for your patience

I suggest you read up on JFETs and what the different specifications mean.    Faster makes no sense to me. 

 

[ricardo]

Mark, please don't be dissuaded trying to make your own mikes.

The Oktava is a really good circuit.  Dorsey's mods only upgrade some bits which Oktava didn't have access to in da old days.  Today, it is very likely that new Oktavas have the very best components to do their job.

If I may comment on a few points.

• the 680p or 820p on the grid of the FET should be a NPO/COG ceramic.  No competent mike designer would use anything else today.  There are loadsa pseudo designers out there
• the 10ns on the XLR pins are for RFI protection.  I used ceramics in the Calrecs and so did Dip Ing Wuttke for Schoeps.  They need to be physically small.  This is for good RF performance.  But the most important thing is that they MUST be directly mounted on the pins of the XLR between p2 & 3 and p1 AND p1 MUST be connected directly to the mike metalwork.  All with the SHORTEST POSSIBLE LEADS.  If they are even 1" from the XLR, you might as well leave them out
• the 10n before the 1G should be a SMALL size polyester
• J305 is a better FET than any 2sk170

Everything else is non-critical.  Just buy good stuff from Digikey & Mouser

 

[mark88]

Mark, please don't be dissuaded trying to make your own mikes.

The Oktava is a really good circuit.  Dorsey's mods only upgrade some bits which Oktava didn't have access to in da old days.  Today, it is very likely that new Oktavas have the very best components to do their job.

If I may comment on a few points.

• the 680p or 820p on the grid of the FET should be a NPO/COG ceramic.  No competent mike designer would use anything else today.  There are loadsa pseudo designers out there
• the 10ns on the XLR pins are for RFI protection.  I used ceramics in the Calrecs and so did Dip Ing Wuttke for Schoeps.  They need to be physically small.  This is for good RF performance.  But the most important thing is that they MUST be directly mounted on the pins of the XLR between p2 & 3 and p1 AND p1 MUST be connected directly to the mike metalwork.  All with the SHORTEST POSSIBLE LEADS.  If they are even 1" from the XLR, you might as well leave them out
• the 10n before the 1G should be a SMALL size polyester
• J305 is a better FET than any 2sk170

Everything else is non-critical.  Just buy good stuff from Digikey & Mouser

thank you.

• the 10n before the 1G should be a SMALL size polyester

-> this would be ok? http://www.digikey.it/product-detail/it/ECQ-E2103KF/EF2103-ND/56420

 

[Rossi]

the 680p or 820p on the grid of the FET should be a NPO/COG ceramic.  No competent mike designer would use anything else today.  There are loadsa pseudo designers out there[/list]

I think you're turning this into a religion. Your statement implies that 90% of todays mic designers are incompetent, because I don't see NP0/C0G being used very often.

 

[ricardo]

I think you're turning this into a religion. Your statement implies that 90% of todays mic designers are incompetent, because I don't see NP0/C0G being used very often.

Very few people know about this polystyrene microphony and even fewer have tested it.  Rossi, you've just joined this very august group .. and I'm not being snide 

There may be mike designers who like microphonic electronics in mikes  ;D  You decide if this is competence.

There are loadsa mikes 'designed' with 2sk170 too.    It may be that these 'designers' never measure the real life noise performance of their mikes.  I fully appreciate that some parts may have Golden Pinnae sound but I prefer eg caps to behave like caps should and get my 'sound' via other means.

Rossi, why don't you listen to the microphony as I suggested earlier.  You at least have taken the trouble to investigate it.  Its a very short step to putting some numbers on this 'feature' and then deciding if

• it will be audible
• you like it

This is in the end, what we are striving for.

Yes.  Electronics that don't add 'sound' to a microphone is a religion for me. 

Having said that, I used a LOT of electronic cheats to get the sound I wanted in the Calrec Soundfields. 

This Millenium, its a lot easier to do this with DSP.  8)

Anyone have a modern Schoeps body and is willing to take it apart?  I'd dearly love to know what Dip Ing Wuttke used in his last designs when he wasn't faced with the constraints we had in the 80's

 

[ricardo]

• the 10n before the 1G should be a SMALL size polyester

-> this would be ok? http://www.digikey.it/product-detail/it/ECQ-E2103KF/EF2103-ND/56420

Yes.  I would personally use 100n from the same series if it would fit in the space.  Voltage only needs to be more than 63V

 

[Rossi]

Very few people know about this polystyrene microphony and even fewer have tested it.  Rossi, you've just joined this very august group .. and I'm not being snide 

There may be mike designers who like microphonic electronics in mikes  ;D  You decide if this is competence.

There are loadsa mikes 'designed' with 2sk170 too.    It may be that these 'designers' never measure the real life noise performance of their mikes.  I fully appreciate that some parts may have Golden Pinnae sound but I prefer eg caps to behave like caps should and get my 'sound' via other means.

Rossi, why don't you listen to the microphony as I suggested earlier.  You at least have taken the trouble to investigate it.  Its a very short step to putting some numbers on this 'feature' and then deciding if

• it will be audible
• you like it

This is in the end, what we are striving for.

Yes.  Electronics that don't add 'sound' to a microphone is a religion for me. 

Anyone have a modern Schoeps body and is willing to take it apart?  I'd dearly love to know what Dip Ing Wuttke used in his last designs when he wasn't faced with the constraints we had in the 80's

Polysterene is not very popular these days, except in vintage/retro designs for period correctness.

Polypropylene is what's mostly used today, and I think it's for sound reasons (pun intended). And I'm talking about real designs by reputable manufacturers, not Chinese contraptions which indeed often use 2SK170 and inferior grade ceramics.

Polypropylene may be slightly more microphonic than C0G, but not enough so to cause trouble. As I've said, in my tests polypropylene sounded cleaner than C0G. It was a bit closer to the plain wire (non-)sound. This may be contrary to what you know but I doubt it's contrary to what you hear, if you do a comparison. Also, this is an application that's not widely tested: extremely high impedance and an uncommonly high voltage differential of about 60 volts. Plus the signal levels can be quite high. Many a large diaphragm capsule puts out around 25 mV/Pa, so we're talking about several volts RMS once we approach the mic's max SPL figure. Those are conditions far beyond the Bateman measurement scenarios.

So while I investigate some more into the microphony issue, why don't you make actual listening comparisons of different input caps, polypropylene and C0G in particular.

The mic input cap is the one area where I don't want any sound coloration. However, I'm not averse to some coloration further down the signal chain, depending on what the mic aims for. There are applications for very clean mics and there are applications for somewhat colored mics. I've built both, and I find it more challenging to build a mic that has a (good) sound of its own, because your goal is a subjective one and it can take a lot of time to refine such a circuit. Clean mic circuits aren't exactly easy, either, but at least you have a clear target to aim for.

On the Schoeps site, there is a pic of the internals of their new V4U vocal mic. It uses a new circuit design which gives higher SPL handling than the CMC circuit. However, I doubt Wuttke designed it, since he retired a few years ago (although he's still involved with Schoeps). As a matter of fact, I'm not sure Jörg Wuttke designed the well known CMC circuit. One of its predecessors, the CMT30 already used a very similar circuit - differences are merely in the DC converter section and a few component values in the actual amplifier). According to www.schoepsclassics.de the CMT series appeared in 1965, which predates the Wuttke era. I don't own a modern day Schoeps, but as far as I know the CMC5/6 still uses the same basic circuit, but I don't know anything about its component choices except that it uses SMD components.

 

[mark88]

Very few people know about this polystyrene microphony and even fewer have tested it.  Rossi, you've just joined this very august group .. and I'm not being snide 

There may be mike designers who like microphonic electronics in mikes  ;D  You decide if this is competence.

There are loadsa mikes 'designed' with 2sk170 too.    It may be that these 'designers' never measure the real life noise performance of their mikes.  I fully appreciate that some parts may have Golden Pinnae sound but I prefer eg caps to behave like caps should and get my 'sound' via other means.

Rossi, why don't you listen to the microphony as I suggested earlier.  You at least have taken the trouble to investigate it.  Its a very short step to putting some numbers on this 'feature' and then deciding if

• it will be audible
• you like it

This is in the end, what we are striving for.

Yes.  Electronics that don't add 'sound' to a microphone is a religion for me. 

Anyone have a modern Schoeps body and is willing to take it apart?  I'd dearly love to know what Dip Ing Wuttke used in his last designs when he wasn't faced with the constraints we had in the 80's

Polysterene is not very popular these days, except in vintage/retro designs for period correctness.

Polypropylene is what's mostly used today, and I think it's for sound reasons (pun intended). And I'm talking about real designs by reputable manufacturers, not Chinese contraptions which indeed often use 2SK170 and inferior grade ceramics.

Polypropylene may be slightly more microphonic than C0G, but not enough so to cause trouble. As I've said, in my tests polypropylene sounded cleaner than C0G. It was a bit closer to the plain wire (non-)sound. This may be contrary to what you know but I doubt it's contrary to what you hear, if you do a comparison. Also, this is an application that's not widely tested: extremely high impedance and an uncommonly high voltage differential of about 60 volts. Plus the signal levels can be quite high. Many a large diaphragm capsule puts out around 25 mV/Pa, so we're talking about several volts RMS once we approach the mic's max SPL figure. Those are conditions far beyond the Bateman measurement scenarios.

So while I investigate some more into the microphony issue, why don't you make actual listening comparisons of different input caps, polypropylene and C0G in particular.

The mic input cap is the one area where I don't want any sound coloration. However, I'm not averse to some coloration further down the signal chain, depending on what the mic aims for. There are applications for very clean mics and there are applications for somewhat colored mics. I've built both, and I find it more challenging to build a mic that has a (good) sound of its own, because your goal is a subjective one and it can take a lot of time to refine such a circuit. Clean mic circuits aren't exactly easy, either, but at least you have a clear target to aim for.

On the Schoeps site, there is a pic of the internals of their new V4U vocal mic. It uses a new circuit design which gives higher SPL handling than the CMC circuit. However, I doubt Wuttke designed it, since he retired a few years ago (although he's still involved with Schoeps). As a matter of fact, I'm not sure Jörg Wuttke designed the well known CMC circuit. One of its predecessors, the CMT30 already used a very similar circuit - differences are merely in the DC converter section and a few component values in the actual amplifier). According to www.schoepsclassics.de the CMT series appeared in 1965, which predates the Wuttke era. I don't own a modern day Schoeps, but as far as I know the CMC5/6 still uses the same basic circuit, but I don't know anything about its component choices except that it uses SMD components.

If I have cleary understand what You and Ricardo said:
-polypropylene caps are cleaner
-polysterene are.. vintage sounding? warmer?
-ceramic C0G/NPO ... do not add coloration to the sound?
Is it right?

• J305 is a better FET than any 2sk170

As for the 2Sk170, there are different versions of the J305? Should I take care of a particular version? Manifacturer? (I've seen Fairchild and Vishay)
thanks

 

[mark88]

• the 10n before the 1G should be a SMALL size polyester

-> this would be ok? http://www.digikey.it/product-detail/it/ECQ-E2103KF/EF2103-ND/56420

Yes.  I would personally use 100n from the same series if it would fit in the space.  Voltage only needs to be more than 63V

this one? (100nf = 0,1uF right?)  http://www.digikey.it/product-detail/it/ECQ-V1104JM/P4725-ND/10746
why a 100nF when the sheet specifies 10nF?

thanks

 

[ricardo]

Polypropylene may be slightly more microphonic than C0G, but not enough so to cause trouble. As I've said, in my tests polypropylene sounded cleaner than C0G. It was a bit closer to the plain wire (non-)sound. This may be contrary to what you know but I doubt it's contrary to what you hear, if you do a comparison. Also, this is an application that's not widely tested: extremely high impedance and an uncommonly high voltage differential of about 60 volts. Plus the signal levels can be quite high. Many a large diaphragm capsule puts out around 25 mV/Pa, so we're talking about several volts RMS once we approach the mic's max SPL figure. Those are conditions far beyond the Bateman measurement scenarios.

So while I investigate some more into the microphony issue, why don't you make actual listening comparisons of different input caps, polypropylene and C0G in particular.

It so happens I HAVE conducted DBLTs on input caps against 'plain wire' (non-) sound ... and also tested microphones at very high spl too.

My religious beliefs are based on these early 1980s tests.

When you do your listening tests on microphonic caps, make sure your test circuit doesn't have anything else that is microphonic as that will reduce the difference between caps.  Just short out the test capacitor and try tapping the mike again.  Actually when you listen to your caps, you will hear stuff quite clearly without having to shout.  Just turn up the gain a bit.

You may have noticed, I suggest simple listening tests for many things.  If I was a guru for anything in my previous life, it was for DBLTs ... but you don't need to go to such lengths to decide if microphonic caps might be making the sound of your mikes worse.

This Millenium, I'm just a beach bum pretending to be a guru on the www 

Thanks for the stuff on Schoeps mikes.  Its actually the detail design I'm after .. eg the type of caps he uses.  I was amused to find Wuttke used exactly the same RF protection as on the Calrec stick mikes though my circuit (transformer) was completely different from his.  I didn't actually see a pukka Schoeps CMC schematic until this century.

The Schoeps CMC schematic on the web is a very old one via Eric.  Do you know of any later ones or of V4U?

why a 100nF when the sheet specifies 10nF?

As you see from above, real mike designers have their idiosyncratic beliefs too.  You'll have to figure out who are the frauds ;D

there are different versions of the J305? Should I take care of a particular version? Manifacturer?

My hands on experience with LN FETs and different makers is from the early 80's.  My favourite FET for mikes was Mullard/Philip BFW11 made by Southampton virgins.  No other FET was in the same league or as consistent.  The Southampton factory is long gone and no modern BFW11 is a LN device.

Today, I would use Fairchild .. simply cos they seem to be the only ones to publish full noise data for their FETs.  The Oktava circuit doesn't really care about Idss.

 

[Rossi]

Mark, you can simply ignore our polypropylene vs. C0G controversy. Either will be fine. Don't focus on the details so much, get some components and try to make it work.

100 nF gives better filtering for the polarization voltage than 10 nF. As you see, sometimes a bigger value is an improvement, sometimes it's worse and sometimes it doesn't make a difference. You have to understand the circuit in order to predict what will happen if you change a component.