Fun and Games Aligning and Servicing a Revox A77 (Dolby B Version)

August - October 2007

I've owned a 1973 vintage Revox A77 (a standard speed, 4 track, Mark 3 with Dolby B) since 1986, when I bought it for £300 from a musical instrument shop called "Tune Inn" in Hither Green (South London). That shop is still there and now has a web site:

I don't think (in retrospect) that this A77 was in very good condition - the heads were worn and the rotating tape guide made a fearful noise when rewinding tape. But I'd wanted a Revox A77 since I saw one at school. That must have been about 1970 and I was 11 years old. The quality of the thing was obvious, though, even at that age. So once I'd seen this one in the shop, I had to buy it!

I used it for recording off the radio on and off, but it never sounded as good as I hoped it would (not suprisingly). Eventually, in 1997, I had some money to spare and I saw an advertisement offering Revox servicing in the back of "HiFi News and Record Review". So - off it went for what I hoped would be a comprehensive renovation.

Unfortunately, that didn't turn out too well ... Everything started off fine - new heads were fitted and all the worn mechanical parts were replaced. Then there was silence for about 3 months ... I think the person doing the work had some sort of personal crisis. Eventually, I managed to get in touch with him, paid him and got the machine back. He'd done a lot of work on it, and mechanically it was now sound. But it still didn't sound all that good and there were definitely electrical problems (audible distortion on one channel and I strongly suspected that it hadn't been set up properly - or at all).

Due to pressure of work, not much happened until 2003, when I fixed the distortion problem (by replacing one of the transistors in the input amplifier). It now sounded a lot better - but still not as good as I thought it should.

Over the years, I'd built up a collection of Hewlett-Packard and Tektronix test gear (thanks mostly to E-Bay) and, having all the necessary equipment to hand, I finally made some time to sort the A77 out once and for all. Or so I hoped ... (I was also prompted by the arrival of audible distortion on the channel I hadn't fixed before!)

"Sorting it out" turned out to be a rather long winded process. One issue was that the Dolby specific aspects are not very fully described anywhere in the three manuals I have to hand. I know that the person doing the renovation in 1997 got really despondent about the Dolby side of things.

These are the manuals I got hold of:

And here is what I did (it goes on a bit - but the Dolby B specific parts might actually be useful to someone!):

1) Clean and demagnetize.

2) Take the wooden case off, so we can get at the innards.

Take to the bench, where we have ...

The order of things from here is determined mostly by the SM section (hidden near the very end) "Alignment Instructions Revox A77 Dolby", which has 25 steps. This doesn't quite cover everything in as much detail as you would like (it is only 2 pages). We'll refer to this as "SMAID".

3) Adjust the 21V DC regulated power supply. (SMAID 1)

-> Read: 21.5V

-> Read: 20.0V +/- 0.01V

-> Read: < 40 mV.

This isn't in the Electrical Adjustments and Measurements part (Section 6) of the SM.

4) Tried to adjust the capstan motor speed ... (SMAID 2,5)

But ended up just measuring it!
See SM 6.2.2.

-> Read: 1596 Hz at 7.5 ips, 801 Hz at 3.75 ips.

Location of the usual (non Dolby) alignment adjustments.

5) Check Bias Oscillator Frequency and Voltage (SMAID 6, SM 6.4.1)

-> Bias oscillator freq was 123.4 kHz (141 Hz SD). Nominal 120 kHz +/- 5%, so OK.
-> Voltage at Orange: 27.0Vrms (32V nominal on 4 track).
-> Voltage at Brown: 13.7Vrms (16V nominal).

6) Adjust Record Amplifier Bias Traps (SMAID 3, SM 6.4.3)

-> Min obtainable was 323 mV (Goal is < 300 mV). Not ideal.

7) Adjust Reproduce Amplifier Bias Traps (SMAID 4, SM 6.4.4)

-> 2mV (Goal < 50mV). This is suspiciously low ... but may be good!

8) DOLBY Record Processor Adjustments (SMAID 7)

This is where the real fun began! Some careful preparation was undertaken to try to make sense of where to measure the many things that needed to be measured. Unfortunately, when making pretty much the first adjustment (to P102), this trimpot literally fell apart! So, there was a pause while the record and playback processors had all their trimpots replaced! This was not so simple. I couldn't find the right size of trimpots at either RadioSpares or Maplin ... Maybe I didn't try hard enough. But a lot of components had recently become "no longer available" due to RoHS compliance requirements... The substitute trimpots were Piher enclosed verticals, and seem to be fine - albeit needing small wire jumpers to attach the legs at either side to the PCBs. Not ideal ... but it seems satisfactory.

To make the required measurements, I found it necessary to tack solder leads from the measuring instruments to the appropriate "pins" on the edge connectors on the "DOLBY Basic Board" (1.077.854).

  1. Attach a "BNC to wires" coax test lead to "P1" (pin 2) with ground to pin 6. This is used to measure AC voltages.
  2. Plug this test lead in to 3400A VTVM or 3457A DMM.
  3. Plug 3325A Generator in to AUX inputs with a "BNC to dual RCA with 50ohm termination" test lead. This lets the accurately calibrated 3325A amplitude settings appear at the RCA plugs.
  4. Set front panel RECORD LEVEL to full.
  5. Set DOLBY switch OUT.
  6. Set trimpots P101 and P102 for CH1 fully anti-clockwise.
  7. Set 3325A frequency to 5kHz and adjust amplitude to get 3mV on 3400A/3457A. This required 0.45mV.
  8. Transfer test lead to "P3" (pin 1). 3400A should now read 2.6mV. It did.
  9. Set DOLBY switch IN.
  10. There should be a +8dB (x2.51) increase (to 7mV approx). There was.
  11. Adjust P102 (CH1) to get a +10dB (x3.16) increase on 2.6mV which is 8.2mV.
  12. Adjust P101 (CH1) to get a -2dB (x0.795) reduction, which is 6.51mV.

  1. Attach a "BNC to wires" coax test lead to "P2" (pin 12) with ground to pin 6.
  2. to g) Repeat these steps for CH2.
  1. Transfer test lead to "P4" (pin 13). 3400A should now read 2.6mV.
  2. to l) Repeat these steps for CH2.

Unfortunately, I noticed that CH2 needed 0.59mV to get the measured 3mV ... indicating a problem with the input amplifier. This fitted with the fact that the CH2 RECORD LEVEL had always had to be set considerably higher than CH1 to get equal levels on the meters and by hearing... More later on this.

9) DOLBY Playback Processor Adjustments (SMAID 8)

These "line up" the Playback Processor to undo the modifcations the Record Processor makes to the signal. So ...

  1. Set P102 (CH1) and P101 (CH1) fully anti-clockwise.
  2. Set DOLBY switch OUT.
  3. Attach FIRST "BNC to wires" coax test lead to "P6" (pin 9) with ground to pin 10. This is used to inject a signal from the 3325A.
  4. Attach SECOND "BNC to wires" coax test lead to "P8" (pin 1), also with ground to pin 10 (or pin 6 on Record Processor if easier). This is used to measure AC voltages and is attached to the 3400A or 3457A.
  5. Adjust 3325A amplitude (at 5kHZ) until 3400A/3457A measures 7.6 mV. Start at about 0.4mV and adjust. It required 3.25mV.
  6. Set DOLBY switch IN.
  7. Adjust P102 to reduce the voltage on "P7" by -10dB (x0.316) to get 2.4 mV.
  8. Adjust P101 to increase the voltage by +2dB (x1.259) to get 3mV.

  1. Attach FIRST (input) lead to "P5" (pin 13).
  2. Attach SECOND (output) lead to "P7" (pin 15).

10) VU Meter Adjustment (SMAID 9)

This is done on the DOLBY Playback Processor board in the DOLBY A77. The "usual" adjustments on the Record Amplifiers do nothing.

  1. Attach a "BNC to wires" coax test lead to "P1" (pin 2) with ground to pin 6 on the Playback Processor edge connector. This is used to measure AC voltages.
  2. Plug this test lead in to 3400A VTVM or 3457A DMM.
  3. Plug 3325A Generator in to AUX inputs with a "BNC to dual RCA with 50ohm termination" test lead. This lets the accurately calibrated 3325A amplitude settings appear at the RCA plugs.
  4. Set front panel RECORD LEVEL to full.
  5. Set DOLBY switch OUT.
  6. Set 3325A to 1kHz frequency and about 13mV amplitude.
  7. Adjust 3325A amplitude to get 100mV measured at "P1" on 3400A/3457A.
  8. Adjust P103 (CH1) on Playback Processor until CH1 VU meter reads 0VU.

  1. Attach measurement lead to "P2" (pin 12)

11) Input Amplifier Repair

There was nothing obviously wrong with the Input Amplifier module, so I started taking out and testing first the electrolytic capacitors, then the tantalums, then all the other passives ... Nothing changed. I then swapped the two PNP transistors over (CH1 -> CH2, etc.). Nothing changed ... which was just as well, as I had no spares. Then I changed all the NPN transistors ... to BC548B's ... and it worked! I did also clean the edge connector with an eraser at the same time. So ... who knows which one fixed it? It seems unlikely that all 4 NPNs were bad - although I had previously changed one in the working channel (CH1) about 5 years ago. That time it was definitely the transistor that fixed a problem with horrendous distortion.

12) Balance Control Adjustment (SMAID 10, SM 6.3)

This is simply the Balance Control.

13) Playback Head Azimuth Adjustment (SMAID 11, SM 6.3.1, 6.3.2)

This is where we adjust the "tilt" of the playback head wrt the tape.

14) Playback Level Adjustment (SMAID 12, SM 6.3.3)

We now set the REPRODUCE LEVELS on the REPRODUCE AMPLIFIERS to get the correct output for a given magnetization on the tape.

15) Playback Frequency Response Check. (SMAID 13, SM 6.3.4)

Using a precorded tape with various tones on it, we observe the playback frequency response.

Freq. CH1 mV CH2 mV
10Hz 138 139
40Hz 300 240
125Hz 210 190
500Hz 174 165
1kHz 154 147
4kHz 139 137
8kHz 135 137
12.5kHz 134 137
16kHz 132 137
18kHz 135 135

Note the low frequency "hump" starting below 1kHz. This is almost certainly normal, and is a result of the test tape having been recorded with 1 full width track. When this is reproduced on a multi-track machine, "fringing effects" known to occur below 1kHz will cause low frequency "bumps", expected to be several dB up (and down) at the bottom end.

16) Record Head Azimuth Adjustment (SMAID 14, SM 6.5.3)

Now we have got the playback side of things sorted out, we can usefully start on the record side!

Now we need to pass the signal through the record-reproduce path:

17) R.F Bias Adjustment (SMAID 15, SM 6.5.1/6.5.3)

The infamous "setting up the bias level for the tape" ... There is a setting of the bias amplitude that will give the best combination of noise floor, distortion and frequency response. This is a compromise - no one bias setting will optimize all three simultaneously! We start at the lowest bias amplitude. Recordings made at this level will tend to sound "bright" and very noticeably distorted. Pretty horrible. As the bias is increased, the high frequency response will first increase (get even "brighter"). Continuing to increase the bias amplitude, the high frequency response will start to fall off, and the distortion and noise will also start to decrease. The "amplitude peak" as the bias amplitude is increased can be seen on a voltmeter. The "best" bias level for a given tape brand is specified as a desired decrease in output relative to the level at the amplitude peak. This desired decrease is tape speed dependent ...

In practice, the "peak" may be very close to the level obtained with minimum bias amplitude, so it can be quite hard to see. But using the level at minimum bias isn't going to be far off under those circumstances.

For a given tape and tape speed, we look up the "best" bias decrease in a table supplied by Revox (SM Table 6.5-69 amended for more modern tapes in the "Supplement to the Adjustment Instructions (Chapter 6)").

For Ampex 456 tape, at 7.5 ips, -6dB (x 0.5) and at 3.75 ips, it is -5dB (x 0.56). With other "modern tapes" (e.g. BASF / EMTEC / RMGI 468) the correct bias settings will be very close to the Ampex 456 levels - or so I believe.

The PEAK LEVEL measurements (in Volts) we actually made doing this were:

7.5 ips Peak Want
CH1 0.43 0.215
3.75 ips Peak Want
CH1 0.36 0.201
CH2 0.46 0.260

Now the good thing is that I can honestly say that these settings also sounded best ... from a quick listen, anyway. The difference between input and recorded output was now very slight.

18) Record Level and Calibration Oscillator Adjustments (SMAID 16)

This procedure is specific to the Dolby version and isn't all that clearly described in any manual. We need to adjust the VU meters in RECORD so that they read the correct level.

Now, we start to record and monitor what comes off tape.


Now we can set the Calibration Oscillator level correctly.

Was it worth it?

Well, yes! It certainly sounds a great deal better now - in fact it IS (finally) as good as I had always hoped it would be. It is quite difficult to hear any difference between "input" and "off-tape" now (there is hiss in quiet passages, of course, but you don't really notice it).

How about some measurements? Here are some frequency response charts. These were made with the HP3325A as a source and the HP3457A as the measuring device, under GPIB control. The graphs were drawn from the automatically measured data using GNUPLOT. These were taken with the record level adjusted to 0VU

Before ...

After ... No filter, No Dolby B

After ... Filter, No Dolby B

After ... No Filter, Dolby B

After ... Filter,Dolby B

Here are some more frequency response plots. These are from the HP3582A. This is an FFT based spectrum analyzer (also known as a Dynamic Signal Analyzer). The input is random noise, with the record level adjusted to -10dB on the VU meters. Please note that these plots have a linear frequency axis.
There are some interesting features about these plots. Puzzles even ... One thing that had me really puzzled for a long time is that the response as shown by this method is quite seriously dependent on the recording level. Doing this at 0VU gives a LOT more high frequency roll off. Is this real ... or what?
Even weirder is that with the Filter ON at 0VU, the response is MUCH FLATTER than with the Filter OFF (up to the frequency where the filter comes in)! I have NO EXPLANATION for this! Unless it ties in with the level dependency. I suspect it is real, as you can HEAR IT! There sounds to be more high frequency present in the white noise (it is white) on playback when recording with the Filter ON that when it is OFF! Very, very unexpected.

HP 3582A, No filter, No Dolby B

HP 3582A, Filter, No Dolby B

HP 3582A, No Filter, Dolby B

HP 3582A, Filter, Dolby B

Finally, I tried to measure the low frequency response of the A77, where I fully expected to see "head bumps" ... quite "bouncy" low frequency response is, apparently, a well-known feature of analog tape machines. There are some very interesting measurements of various studio tape machines at this web site. BUT ... I measured nothing like this... Another puzzle, since I would expect the relatively humble A77 to be worse than the studio machines that were measured. I think my measurement is wrong. But why?

HP 3582A, Filter, Dolby B

All in all - quite fascinating! (If you like this sort of thing).