Bias Capacitor?

[iomegaman]

I'm doing Jim WIlliams upgrade on my Tascam 38 reel2reel and in it he suggest changing out the bias cap to "A 220 pf/1000V Wima FKP-2 does the bias trick for +9"...in the schematic I am assuming it is C104 which is currently a 270pf (those old green flat turtle looking ceramics)...

My question is, will a slightly lower farad cap but much newer Wima make that much difference in boosting the bias to get that extra +9 and if so, how does it do it...is it a matter of a faster recovery time (the Wima is 1000v so really what time is it?)?

Or I am reading the schematic wrong and not understanding what Jim has suggested?

By the way the primary mod that I am doing is replacing the JRC7002 (U101) with LME49720/changing C102 from a 100uf/10 to a 470uf/25 taking out the input tranx and jumping 1-3 changing out resistor R115 from 2.2k to a 91 Ohm (input leg 2 of opamp )...besides going through the psu and upgrading caps...I am also considering adding a direct wire ground to each card to star instead of simply using the ribbon cables cascaded to ground for each card to cut some cross-talk...

Oh yes I also added two .1uf/100v off leg 4/8 of opamp power rails to prevent oscillation...

 

[ruffrecords]

I don't think it is C104, which is in the NFB loop of the replay op amp. It's purpose seems to be to ensure the response drops off above 100KHz.

I would have expected the capacitor concerned to be somewhere in the bias amplifier circuit.

Cheers

Ian

 

[iomegaman]

Yes thanks Ian found the bias part of schematic and the current one is apparently adjustable...gonna track down the rest of the bits...

 

[abbey road d enfer]

I'm doing Jim WIlliams upgrade on my Tascam 38 reel2reel and in it he suggest changing out the bias cap to "A 220 pf/1000V Wima FKP-2 does the bias trick for +9"...in the schematic I am assuming it is C104 which is currently a 270pf (those old green flat turtle looking ceramics)...

My question is, will a slightly lower farad cap but much newer Wima make that much difference in boosting the bias to get that extra +9 and if so, how does it do it...is it a matter of a faster recovery time (the Wima is 1000v so really what time is it?)?

Or I am reading the schematic wrong and not understanding what Jim has suggested?

Can you be a little more specific on what components are changed for whar expected result? Modifying the bias circuit in order to  increase its amplitude is heretic. The bias has an optimum value that is determined by the tape formulation and the construction of the heads; increasing bias blindly just conducts to treble loss and increased THD.

By the way the primary mod that I am doing is replacing the JRC7002 (U101) with LME49720/changing C102 from a 100uf/10 to a 470uf/25 taking out the input tranx and jumping 1-3 changing out resistor R115 from 2.2k to a 91 Ohm (input leg 2 of opamp )...

Again, what is the expected benefit? Bypassing the transformer would be a nice move if it was not plagued by other consequences, much worse IMO. The transformer is used to bring the source impedance of the head to a value that is close enough to the OSI of the opamp - Optimum Source Impedance. Without the transformer the opamp will have a stupid noise factor. The LME49720 is indeed better than the JRC4558 in terms of noise figure but the impedance mismatch is such that the actual noise performance is 8dB worse . 

  I am also considering adding a direct wire ground to each card to star instead of simply using the ribbon cables cascaded to ground for each card to cut some cross-talk...

Do you really have a x-talk problem that you can trace to ground "unstiffness"? Generally, x-talk in tape recorders is limited by the heads. Adding a star ground is not the best way to improve x-talk anyway, far from it.

Oh yes I also added two .1uf/100v off leg 4/8 of opamp power rails to prevent oscillation...

This looks to me the only sensible action so far...as long as they are connected to a suitable "ground".

 

[iomegaman]

Sorry, yes I have decided NOT to change any capacitor values to modify bias and agree with your assertions...the opamp replaced is the JRC 4558DD and in my limited experience I do find them a bit "murky" but as all opamps are subjective to the rest of the signal path it is purely my opinion...

The tranx is a little Tamura line input that gets the input impedance (via a 2.2k resistor) to 2.5k

Here's the original quote from Tape Op where Jim makes his suggestions:

If you want to get into it, you can replace the opamps with National LME49720's or AD8599's and then pull out the cheesy step up transformer used to get cheap gain off the heads. Bridge it and the remove the resistor from pin 2 to ground on the opamp. Replace it with a 91 ohm or so, low enough to make up the lost gain to unity. That will open up the playback quality to no end. They can then sound like MCI -JH-24 or Studer A-800. It's also possible to replace a couple of parts in the bias circuit so you get more, enough for +6, maybe even enough for +9 levels with GP-9 type tape. The output trims can be elevated to get +4 unbalanced outputs and reducing the record trim allows +4 db inputs as well. Lots can be done to tweeze these machines, I made a few very good sounding records with one back in the early 80's on a heavily modified machine. Doug Sax at Mastering Lab was my mastering guy then and he always commented on the excellent sound I brought and was very suprised to hear they were recorded on a modified 38 and a custom console.

 

[abbey road d enfer]

The tranx is a little Tamura line input that gets the input impedance (via a 2.2k resistor) to 2.5k

I don't know how you come to this conclusion...the transformer has a ratio of 1:8, so whatever impedance presented at the primary is reflected with a factor 64 at the secondary. The head's impedance is inductive so varies largely with frequency. I don't know the head's specs but I guess it's about 1mH, 64mH reflected, which computes at 800 ohms at 1kHz, 8kohm at 10kHz, 16 kohm at 20kHz. That fits with the OSI of about 16kohm (8nV/sqrtHz and 0.5pA/sqrtHz).

Here's the original quote from Tape Op where Jim makes his suggestions:

If you want to get into it, you can replace the opamps with National LME49720's or AD8599's and then pull out the cheesy step up transformer used to get cheap gain off the heads. Bridge it and the remove the resistor from pin 2 to ground on the opamp. Replace it with a 91 ohm or so, low enough to make up the lost gain to unity. That will open up the playback quality to no end. 

JW's analysis of the role of the input transformer is flawed. He surmises that it's there just to provide "cheap" gain. But it's there primarily to present the 4558 with the right source impedance.
In the original arrangement, the wide-band input noise is about 1.5uV. With the new, transformerless arrangement, the input noise is about 0.38uV, or 12dB less  BUT the overall gain must be increased by 18dB in order to compensate for the 1:8 transformer gain. As a result, the noise performance is 6 dB worse.

 

[iomegaman]

Thanks for that, JW always defaults to changing out opamps for the flavor he prefers...sometimes I suppose it's best to leave things where the engineers designed them.

 

[joaquins]

Sorry, yes I have decided NOT to change any capacitor values to modify bias and agree with your assertions...the opamp replaced is the JRC 4558DD and in my limited experience I do find them a bit "murky" but as all opamps are subjective to the rest of the signal path it is purely my opinion...
...

Keep those JRC opamps, are great for guitar overdrives. They tend to have more distortion than other 4558 opamps, but works really nice in guitar overdrives, this exact device was used in the TS9 which is a nice one, I build one and the difference with other opamps is noticeable.

JS

 

[iomegaman]

Abbey road d enfer...please forgive my lack of understanding but in your equation I get the noise floor issue, but this brings up a question of design for me...

what happens if you put a transformer back into the circuit but AFTER all the opamps? suppose you added a 1:8 output tranx right before the unbalanced outs? And maybe of course there is no room for such a beast that's why the small footprint one was on the board to begin with, but in actual practice wouldn't an output tranx with the lower STN make more sense?

 

[PRR]

> put a transformer back into the circuit but AFTER all the opamps?

Has no effect on the hiss. Hiss is all in the first stages.

"Cheap gain" may be confusing you. The cheapest gain, by far, is a '4558. One '558 can give gain of 50 without sucking, two '558 stages cascaded can give gain of 2,500, all for 19 cents. That "cheezy" transformer is only gain of 8, and musta cost several dollars.

The transformer IS valid when you have a pre-choosen source (a head) which is hundred-Ohm impedance, and the only amplifier you have gives best hiss at several-KOhm impedance. A low-Z head is easier to wind (important when you jam 8 in a small space) and less trouble with capacitance (in the head, and in the interconnect to the play preamp). Historically, affordable chip-amps gave best hiss for many-K impedances. Tascam seemed to own (or partner with) a transformer winder, so they went that path.

(There are other reasons why you'd put a transformer on the OUTput. It is still the gold standard for Balanced/Floating Out. It can step-up a 24V supply for >24V p-p audio output. But it does nothing for the hiss. And a tranny to carry >10V costs MUCH more than one to carry 10mV.)

I think Jim is recalling old work and not remembering (or revealing) all that he knew then. I'm sure he'd be pleased if other people Used Their Brains and thought-about what might improve the '38. He's moved on and probably won't write-up a Paint By Numbers Dummy-Guide for this obscure project, even if he did remember every little detail.

The Tascam circuits are usually good bones, but some were a little ahead of good *and* cheap chips. The 4558 was Tascam's jellybean, and the key to their prices and popularity. Today the 5532 costs a penny more, is for-sure good sound, and may be a little better in many Tascam holes. Specifically for the head preamp, I would verify the transformer ratio, and preferably head impedance, then search for a suitable chip. If the numbers above are close, a Jensen 990 might be an excellent head preamp, though an 8-pack costs far more than a '38 is worth. There are several low-Z chips today with fine audio performance at reasonable prices. But it isn't "my" Tascam 38 so I am not inclined to do the research for you.

 

[iomegaman]

Thanks PRR good info as always, why I love this site, no one lets you get away with being lazy which is a good thing in audio and life in general.

This is my first foray into Tape, I picked up this Tascam 38 with the 2 noise reduction modules for almost nothing but the trip across town (I traded some woodworking materials I will never use for it)...so I spent a week researching what others had done how it sounded, blah blah blah...found 3 threads where Jim piped in this same exact info on TapeOp/Gsluts and also found several people like myself who misunderstood all that was involved...so I have been educating myself and currently have spent a lot more money than if I had bought a quiet pet...(reference MRL tape is PRICEY, and 1/2" tape is as well)...demagnetizers, rubber conditioner/etc...it seems like a fun experience where I will get wisdom and not much else.

I had a small pile of LME49720's in a drawer and thought I'd give his mod a go...currently I am waiting on a belt...from down the ebay...

I did get a copy of the manual and I must say one of the great things about old gear is the documentation...(blurry as hell but thorough as hell as well)...like my old Sony MXP21 the documentation borders on a work of art...oh for the old days when secrets were open to anyone brave enough to compete.

 

[PRR]

Actually--- here's a 2-bit way to make gain of 8 and OSI around 100 Ohms.

Replace the "cheezy" input iron with a high-current BJT.

Signal level here is so incredibly low, ~~2mV at U101 input, that distortion is a non-issue.

I have not even simmed it. I'm sure it will bias-up and give gain. I am fairly sure the noise voltage will be super-low (especially for the cost).

Build one with a socket, buy many 2N4401, and sort for hiss. 99 out of 100 2N4401s will have the same low hiss, but you could luck-out with a hisser, toss it. For the final build, don't socket, that's begging for bad contact when the musicians get hot. (Though as this is the play side, maybe not a disaster.)

Depending on power supply come-up, it may "pop" at turn-on. Cover your ears or mute the monitors.

 

[iomegaman]

THANKS! I will definitely give that a shot (and let you know how it sounds) I've got some sockets and probably a handful of 2N4401's about...these Playback/Rec boards slip out and are pretty easy to mod about on...since there are 8 of them I may try several options and just compare the results...

I'm trying to come up with a reasonable method of using the exact same waveform (not simply a test tone but actual music or something) to kind of A/B a lot of things in my shop...I wonder if leaving on my Lexicon Jamman for few hours with an audio loop is expecting to much...I suppose a test tone tells all I need to know though.

thanks again, wisdom or money...in this case wisdom.