Sanity Check - Convert 1/2" 4 Track to 2 Track (With a twist)

[abbey road d enfer]

The clipping occurs in the rec preamp stage around Q4, as you stated. That's the bottleneck. When that stage clips, I'm still a good 2-3dB shy of the 1% 3rd harmonic spec on modern tape (using ATR).

What voltage is it at the output of Q4? Is clipping symetrical?

I didn't realize just how low the input impedance of the main record amplifier was until I took a closer look after you mentioned it. I'm intrigued, because most of the theory I'm comfortable with is with high impedance inputs.

At the times it's been designed, many engineers had not made the transition from tubes to SS; maybe they were more comfortable with RF concepts, such as common-grid (that translates into common-base) for HF performance...?

You are also correct in regards to the placement of the VU, it's literally strapped right across the secondary of T2 (which is 1:1), right after the line amp.

Well, there's room for improvement here; that would not impact the quality of recordings though. Just monitoring and playback.

As an aside... I'm 31 years old. My first studio experience as a teenager was with blackface ADATs. I missed the entire era of analog tape, but have a tremendous fascination and respect for the great many people who pioneered and cultivated the craft of recording on this medium. It appears to be an era in which recording *engineer* really meant something. I have no misconceptions that tape is magic, and I'm not chasing a particular sound. But I do find modern recording tools make me feel very detached from the engineering side of things. It's too easy I think. Knowing you're working with devices and tools that real people, with real hands, who take real pride in their work, spent countless hours creating (as opposed to an assembly line in china) has an impact on the quality of work you do. At least it does for me.

I made an abrupt transition from tape to digital, fortunately escaping the whole folly of DAT, BVU editing, 8"-floppy-based automation; that's because I didn't have to make a living on this.
I'm happy to be free of constantly watching needles, and yet having hiss issues, weird kick-drum sound, pesting about the missing 25th track and the flanging sound of tracks 1 & 24.
OTOH, I'm always in awe when I see a tape machine, equally for a Studer A8xx (the pinnacle of no-compromise design) as well as a 1970's Grundig or Philips (how to build a workhorse at a price).

 

[abbey road d enfer]

The clipping occurs in the rec preamp stage around Q4, as you stated. That's the bottleneck.

I've simulated both circuits and it shows that the record amp should be capable of running at +15dB above operating level. That would be true below 1kHz; at higher frequencies, there may be clipping, but it's not an issue since recording HF at any elevated level results in self-erasing. That would mean driving the record amp at about +12dBu (3Vrms)
Now the Record preamp is perfectly capable of delivering the required voltage, so there may be something we don't see...
Can you check the levels at R101 (Rec level) and at Q4? Do you have an o'scope?

 

[bjoneson]

I do have a scope, and will take some more measurements tonight. As I recall, when I set the record level pot (500 ohm R101) to achieve 250nw/m at 1kHz, Q4 begins to clip 10dB above that. The bottom rail (0V) clips first. This is while supplying +4dBu RMS to the input (which is a 15000:600 transformer not shown, who's secondary is strapped across R101)

ATR Tape claims 3% 3rd harmonic at 16.5dB over 185nW/m (which would be 13.5dB over 250nW/m).

In summary, when everything is set to print 1kHz at 250nW/m, there is only 10dB of headroom remaining in the record preamp before clipping.

Will post some scope shots later tonight.

Many thanks,

Bob

 

[bjoneson]

Did some measurements, and things don't look nearly as bad as I recall, but still potential for improvement I think.

Set up the replay electronics using a 250nW/m test tape, and recorded 1kHz at +4dBu, adjusting record level (R101) until output level was matched. Ramped up from there, measuring output of rec preamp at Q4. Clipping began just over +18dBu (14dB over reference) due to hitting the bottom rail.

Ideally would like the electronics to accept signals at least to +20dBu (16dB over reference) without clipping with the machine set up for 250nW/m reference fluxivity.

I did confirm the +10.5V bias voltage at the base of Q4 (with no input signal). There seems to be additional room on the positive rail (24V) Perhaps increasing the bias voltage closer to 12V would maximize the available swing?

 

[abbey road d enfer]

Did some measurements, and things don't look nearly as bad as I recall, but still potential for improvement I think.

Set up the replay electronics using a 250nW/m test tape, and recorded 1kHz at +4dBu, adjusting record level (R101) until output level was matched. Ramped up from there, measuring output of rec preamp at Q4. Clipping began just over +18dBu (14dB over reference) due to hitting the bottom rail.

OK, that confirms the results of simulation.

Ideally would like the electronics to accept signals at least to +20dBu (16dB over reference) without clipping with the machine set up for 250nW/m reference fluxivity.

I think it's gonna be hard to do in view of the limited power supply rail, with the current structure.

I did confirm the +10.5V bias voltage at the base of Q4 (with no input signal). There seems to be additional room on the positive rail (24V) Perhaps increasing the bias voltage closer to 12V would maximize the available swing?

Positive peaks are almost unlimited because the transistor can nearly hit the + rail. Negative peaks are limited by the emitter resistors. Good practice would set the emitter voltage somewhat over half-voltage. I would say about13-14V. But don't expect miracles. That may earn about 1-2dB.
The problem with this stage is that operating point and gain are interdependant. I guess you could improve by reducing R9 from 100 to 91r or even 82r. That would simultaneously increase headroom AND gain.

 

[bjoneson]

The problem with this stage is that operating point and gain are interdependant. I guess you could improve by reducing R9 from 100 to 91r or even 82r. That would simultaneously increase headroom AND gain.

I got curious and simulated this myself (something I've never done before). You are correct, 82 Ohms appears near optimal to maximize gain and available headroom in the record preamp block. 1-2 dB may be all I need to hit the target. By the sim, may actually be able to squeeze 3dB more out of it. I really can't thank you enough for helping to point me in the right direction on this!

 

[bjoneson]

Spent the last several weeks watching copious amounts of YouTube videos on BJT operation, and learning a circuit sim tool.

Bottom line I've been trying to achieve is how to print more flux to tape before the record amplifiers clip.

I still don't  have a 100% grasp on *how* the record amp functions, but I do now understand it to be a voltage controlled current source. Which makes sense given the rising impedance of the inductive record head.

As I was fiddling around in the circuit sim, trying to achieve more gain out of the record amplifiers block, I noticed something interesting about R26. It's a 5k reostat, marked for "low frequency rec premphasis"

This premphasis was part of the NAB standard intended to reduce mains hum on playback. I use IEC EQ, which does not use this preemphasis.

The EQ cards for IEC, short C9. I assume when the cap is in, it creates an RC filter with R26, and thus the preemphasis.

In any case, I do recognize that this is working as feedback to the first transistor stage, Q6. Depending on the wiper position, the feedback can be significant, essentially creating a short from the emitter of Q7 to the base of Q6 for signal (it's  AC coupled via C8).

In a way, R26 is acting as a gain control for this block. Sure enough, when I checked my boards, R26 was turned fully to the position that would provide maximum feedback.

By turning in the wiper back the other direction, introducing resistance to that feedback path, there is gobs of gain to be had. In fact, in the sim, I found I could actually increase R26 to 10k to achieve even more gain before things started breaking down.

In practice, that little pot yielded gobs of additional headroom I was looking for in the electronics.

There is no mention of this pot anywhere in any of the Scully documentation.

In any case, thought I would share the finding!