Roland JC-120 mod

[Sparrows78]

Hi all!
I have aquired an old Roland JC-120 revision 4 from 1980 that has a pronounced hiss and ridiculously loud volume . From what I have read this is normal for these amps.
I have been thinking and testing and realised that the hiss is coming from the preamps and there is also a "swoosh" coming from the chorus when activated.
This "swoosh" is constant no matter the volume of the amp and therefore more noticable at low volumes. 
On my particular model you have two "MAIN IN" inputs that bypasses the preamps, allowing you to use an external preamp in stereo. When using a modern preamp (Headrush Gigboard in my case) there is no hiss and the volume is also more manageable.
So, I tried putting volume pots at this insert points and that lowered the preamps output volume and it also lowered the hiss and swoosh.
This, of course, resulted in lower overall volume of the amp but that is only benificial in my situation. It is still loud enough.
So I like this "solution", but how to implement it in a way that does minimal change to the amp?

When looking at the schematic I found that R33 (x2 there is one for each channel) is part of a voltage divider for each preamp output. I tried a lower value of R33 and it works! Lower volume and much lower noise level.
My question is if this is the best way of doing this? Does it change impedance in way that may change the response of the power amps? To my ears it sounds fine but I imagine there is less bass than original, is this possible?

I attach schematic with R33 encircled.

[Rob Strand]

My question is if this is the best way of doing this? Does it change impedance in way that may change the response of the power amps? To my ears it sounds fine but I imagine there is less bass than original, is this possible?

There's no right place.  The right solution is to use log taper pots for the volume pots instead of the existing linear taper pots.   That wouldn't change anything at all other than improve usability.

What you have done is fine in small doses.

If you have change the 100k divider resistor to too small a value that means you would have to increase the volume at the volume pot for the same loudness.   That would risk clipping the two stages after the volume pot earlier than normal, despite the amp being perceived at the same level as before.

There's no 100% benign mod you can do around the volume pots to reduce the level there either.

[Sparrows78]

I see what you mean. But the problem is that most of the noise is coming from the gainstages AFTER the volume pot, there is noise even with the volume at 0. And the preamp gain is so high that you can barely use the amp volume pot above 1 in a home recording situation.
Also, the "swoosh" from the chorus is being introduced after the volume pot, right before R33.

Is there any problems using a small number R33, other than the risk of overdriving the preamp?

[Rob Strand]

I see what you mean. But the problem is that most of the noise is coming from the gainstages AFTER the volume pot, there is noise even with the volume at 0. And the preamp gain is so high that you can barely use the amp volume pot above 1 in a home recording situation.
Also, the "swoosh" from the chorus is being introduced after the volume pot, right before R33.

Reducing R33 will help with noise.   Reducing R33 reduces *both* the noise and the signal at R33. For a given perceived amp level a lower R33 requires the volume pot setting to be increased.   That increases the level of the *signal* going into the chorus and as a result the signal to noise improves.  As mentioned before the downside is risk of overload.   Finding a balance between overload and noise is pretty much the whole balancing act of audio systems.    (I should also point out that changing the pot to different tapers will not improve the signal to noise.)

Is there any problems using a small number R33, other than the risk of overdriving the preamp?

It should be OK.

The change in the LF cut-off you mentioned earlier is pretty small.  The -3dB point is about 16Hz normally which would move upto 23Hz with a small R33.

[Sparrows78]

Reducing R33 will help with noise.  Reducing R33 reduces *both* the noise and the signal at R33. For a given perceived amp level a lower R33 requires the volume pot setting to be increased.  That increases the level of the *signal* going into the chorus and as a result the signal to noise improves.  As mentioned before the downside is risk of overload.  Finding a balance between overload and noise is pretty much the whole balancing act of audio systems.    (I should also point out that changing the pot to different tapers will not improve the signal to noise.)

Great, thanks for your input and help, I will try out different values of R33 to see what works best for my situation.

The change in the LF cut-off you mentioned earlier is pretty small.  The -3dB point is about 16Hz normally which would move upto 23Hz with a small R33.

How does R33 impact the frequency corner of a high pass filter in this case? Isn't it only C31 and R61 that forms a high pass(the two components following R33)?

[antonis]

The change in the LF cut-off you mentioned earlier is pretty small.  The -3dB point is about 16Hz normally which would move upto 23Hz with a small R33.

How does R33 impact the frequency corner of a high pass filter in this case? Isn't it only C31 and R61 that forms a high pass(the two components following R33)?

It's a bit more complex..
C31 sees R61 & Q11 input impedance on one side and R33, R32 & R29 on its other side..

[Rob Strand]

How does R33 impact the frequency corner of a high pass filter in this case? Isn't it only C31 and R61 that forms a high pass(the two components following R33)?

When you learn about filters you see C31 and R61 and see a high-pass filter.  Then you learn the -3dB point is f3 = 1/(2*pi*R*C) = 1/(2*pi*R61*C31).

All well and good!

The things is, that only works for a simple circuit.  A circuit where:
- the source impedance is low
- the load impedance is high.

When you have a load impedance most people won't have a problem dealing with a load impedance.  You just make R the parallel combination of the filter resistor and the load resistor.

Where people get this stuff wrong is then you have a source impedance.  When you look at the circuit your eyes still see a nice high-pass filter form.  But from the circuit's perspective the source impedance affects the circuit.    In short you need to use R as the series combination of the source resistance and the filter resistor.

For the Roland circuit the source impedance isn't so easily seen because the source impedance is net combination of R29, R31, R33.  So we need to work that out first.  If you want to read-up on something look at Thevenin impedance.

The first three figures work out the source impedance of the divider R29, R31, R33.

The fourth figure is what the circuit/filter sees.

The fifth figure shows how you can bend things to see why you need to incorporate the source impedance into the filter.  In other words the cap sees the two resistors in series.  When you do that the high-pass filter cut-off is f3= 1/(2*pi*C31*(R1+R33)).  Overall you have a high-pass filter and a voltage divider.



As far as modifying R33 goes.  With R33 at 100k the filter sees 33.3k+68k=101.3k.  However if you made R33 very low then the filter sees 68k.  The smaller value (68k vs 101.3k) pushes the high-pass cut-off up, but not by a lot.

[antonis]

Shouldn't we also put R23 (Q4 Drain resistor) up on the game, despite its compartively low value..??
(just to make OP hate us furthermore..)

[Sparrows78]

For the Roland circuit the source impedance isn't so easily seen because the source impedance is net combination of R29, R31, R33.  So we need to work that out first.  If you want to read-up on something look at Thevenin impedance.

I will definitely read up on this. I am only a hobbyist with very basic understanding, but I'm eager to learn more. I really appreciate the time and effort you put in to help me   

[Sparrows78]

Shouldn't we also put R23 (Q4 Drain resistor) up on the game, despite its compartively low value..??
(just to make OP hate us furthermore..)

Keep it coming! 

[antonis]

I am only a hobbyist with very basic understanding, but I'm eager to learn more.

Consider a simple RC HPF consisting of a series cap and a shunt resistor, driving an infinite value load and driven from a zero impedance source..
Resistor's lower leg (the grounded one) should be considered shorted with source's output (both at AC ground) hence R & C set in parallel..

Then, deviate from ideal conditions (infinite load and zero impedance) and make proper Thevenin calculations..

P.S.
A common "mistake" widely made is the CE amp output HPF calculation when for corner frequency calculation is taken into account only output cap and volume/level pot value..
(which is usually OK for Collector resistor values much lower than volume pot ones but it isn't technically correct..)

[Sparrows78]

So, while trying out different values of R33 I noticed two issues.

The first issue is that on my version there is a crude built-in noise gate when chorus is activated.
It works like this: if output at channel 1 speaker falls below 100mV the chorus shuts of. In the schematic this is shown as a connection "2" at EXT-SPEAKER. This connects at "2 1 CH 20", shown in lower right corner of the schematic. How this works is beyond my understanding. Is there a way to remove this gate all together? When lowering R33 The output is also lowered, wich in turn gates the chorus too early.

The second issue is that the preamp channel 2 is distorting too early. The signal distorts at Q4 output (C12). Before this point the signal is clean even at full volume. In my particular unit R23 is removed and R24 is replaced with a jumper. Can this affect the bias in a way that makes Q4 (25K117 GR) distort?



     

[Rob Strand]

The first issue is that on my version there is a crude built-in noise gate when chorus is activated.
It works like this: if output at channel 1 speaker falls below 100mV the chorus shuts of. In the schematic this is shown as a connection "2" at EXT-SPEAKER. This connects at "2 1 CH 20", shown in lower right corner of the schematic. How this works is beyond my understanding. Is there a way to remove this gate all together? When lowering R33 The output is also lowered, wich in turn gates the chorus too early.

So the easiest mod I see is to short the collector and emitter of Q25.  That will disable the noise gate JFET - in a way similar to the original ckt works with a signal present.

The second issue is that the preamp channel 2 is distorting too early. The signal distorts at Q4 output (C12). Before this point the signal is clean even at full volume. In my particular unit R23 is removed and R24 is replaced with a jumper. Can this affect the bias in a way that makes Q4 (25K117 GR) distort?

Something not clear on the original schematic is the power supply.  The +15V rail doesn't make sense.  And because of this oddity I'm not sure what voltage the +27V and +15V rails really are on your unit.

If the supply to the preamp JFETs is 27V then the R24, R23 values on the schematic make more sense for best headroom.

If the supply to the preamp JFET is actually 15V due to some screw-up by Roland then then values in you unit make sense.

If you measure the voltage on either side of the drain resistor of the preamp JFET (say Q4, R26 and Q3 R17).   A better decision could be made.

And yes, if the biasing is off it could reduce headroom.   However, improving the head room may be changing from the intended sound.   Maybe we need to look at some later/different versions of the JC units to work out what the intent is.   Roland sometimes release mod or change documents when they have errors in built units.

[Sparrows78]

So the easiest mod I see is to short the collector and emitter of Q25.  That will disable the noise gate JFET - in a way similar to the original ckt works with a signal present.

Is it safe to short the collector and emitter? I'm not questioning your knowledge but I read that it can damage the transistor or other components. And just out of curiousity, wich one is the noise gate JFET? I never got my head around the noise gate circuit.

If you measure the voltage on either side of the drain resistor of the preamp JFET (say Q4, R26 and Q3 R17).  A better decision could be made.

Q4 R26 measures 26,6V and 16,8V
Q3 R17 measures 19,2V and 15,6V

worth mentioning is also that on my unit R19(looks a bit like R14 on the schematic) is replaced with 10K(original is 22K).

[Rob Strand]

Is it safe to short the collector and emitter? I'm not questioning your knowledge but I read that it can damage the transistor or other components. And just out of curiousity, wich one is the noise gate JFET? I never got my head around the noise gate circuit

It's good to questions things.  The pic below shows the reasoning behind it:

Q4 R26 measures 26,6V and 16,8V
Q3 R17 measures 19,2V and 15,6V

worth mentioning is also that on my unit R19(looks a bit like R14 on the schematic) is replaced with 10K(original is 22K)

Hmm, I see the problem. There's two R17's and two R19 etc depending on the channel.  Q3 on channel 1 looks like it is powered from 27V and has R17=15k.  Q3 on channel 2 is powered from maybe 19V and ha R17=10k.  Duplicate designators are really asking for trouble if you are making mods!!!

Also notice that the part values between the channels are different values for the same designator.

The main point is I think we can conclude the main power rail to the preamp(s) is a nominal 27V.

I'm not actually sure what channel you measured the voltages for.   From my end it's pretty confusing.  I'm not sure what values are used on what channel in the actual unit.  So I can't judge the biasing very well.

A few posts back you mentioned the unit had different part values to the schematic.  Are you sure those values were for channel 1?

[Sparrows78]

Hmm, I see the problem. There's two R17's and two R19 etc depending on the channel.  Q3 on channel 1 looks like it is powered from 27V and has R17=15k.  Q3 on channel 2 is powered from maybe 19V and ha R17=10k.  Duplicate designators are really asking for trouble if you are making mods!!!

Also notice that the part values between the channels are different values for the same designator.

The main point is I think we can conclude the main power rail to the preamp(s) is a nominal 27V.

I'm not actually sure what channel you measured the voltages for.  From my end it's pretty confusing.  I'm not sure what values are used on what channel in the actual unit.  So I can't judge the biasing very well.

A few posts back you mentioned the unit had different part values to the schematic.  Are you sure those values were for channel 1?

Yes, it is confusing! I'm sorry that I wasn't clear, the measurements are on Channel 2. Channel 1 Is working fine and has no different parts values. Now, I also took some measurements on channel 1 and to sum it up:

Channel 1
Q3 R17: 34,1V and 21,1V
Q4 R26: 34,1V and 18,3V

Channel 2
Q3 R17 19,2V and 15,6V (with R19 replaced with 10K)
Q4 R26 26,6V and 16,8V

To make it even more confusing Roland decided to make discrete voltage +27V and +43V rails for each channel/pcb).
When measuring the separate rails I got these results:

Channel 1
+33,9V (@27V rail) and 39,8V (@43V rail)

Channel 2
+26,7V (@27V rail) and +45V (@43V rail)

[Rob Strand]

Yes, it is confusing! I'm sorry that I wasn't clear, the measurements are on Channel 2. Channel 1 Is working fine and has no different parts values. Now, I also took some measurements on channel 1 and to sum it up:

That helps clear things up quite a bit.

The issue I see I channel 2 is running the JFETs at lower supply voltages and running possibly them at higher gain.   Not the best way to maintain the headroom.  The seems to have done that on purpose to get some dirt on channel 2.

I'll have to ponder what to do.   Channel 1 has the bias dividers on the gates which helps set the gain and optimal bias points independently.   The Q3 stage of channel 2 does not.  You can increase the supply voltage on Q3 channel to by decreasing R19 (sch 22k, unit 10k) to 1k. For Q4 you can increase R27 (470R schem) to decrease the gain then tweak R24 (was 10k on schem).     It really needs a whole redesign.   The lower supply voltage will impose more limits on ch2 than ch1.  Also if you drop the gain is ch2 too much to avoid clipping the gain might contrast too much with ch1.

I'm not really switched on enough today to do it.  Also, I'm not sure how things are will go for the next couple of days.

[Sparrows78]

I'm not really switched on enough today to do it.  Also, I'm not sure how things are will go for the next couple of days.

No problem, you have already helped me more than I would ever think a stranger would on a forum 

I will try out different values and see how the amp respond.

[Sparrows78]

Btw this mod worked flawlessly. No more lost chorus at low volume 

Thanks again for the help!

[Rob Strand]

Btw this mod worked flawlessly. No more lost chorus at low volume

Cool, no reason it should have worked.

Regarding channel 2 overload.   The following changes should help prevent Q4 overloading on channel 2.  It will reduce gain on Q4 Channel 2 (by about 6dB).

Before starting:
* Check R25 is in place, and is 100k

If that checks out:
* Change R27 to 1.2k
* Install 1M into position R23
* Remove short from R24 and
  Install 22k into R24.

You will have to judge whether that's too much or too little gain reduction.

I left Q3 alone since earlier you mentioned it was Q4 overloading.