Bravo 12AU7 Mods and Appreciation thread

[oohms]

Amidst this Bravo/Indeed headphone amp craze, i ended up buying the 12AU7 version, when most people ended up buying the 6922 version ;D

So.. i thought i would start this thread dedicated to modding the 12AU7 version.


First off, some photos:

Topside view


Bottom view



Now for the schematics
(I drew both channels so it becomes easier to see what is shared between the 2 channels)
Ignore the 12V label, it runs at 24V just like the 6922 version




Mods:

Mosfet replacement mod
Replacing the stock mosfet with one of a lower capacitance will make the highs less rolled off. If there is one mod you do to your bravo, this should be it!
The stock IRF630 causes a noticable rolloff, but replacing it with an IRF510 removes this, and brings back the treble.


Power supply filtering
If you want to stick with a switchmode power supply, adding an inductor in series with the supply voltage (before the big filtering capacitor) the switching noise should be reduced.
One step further would be to upgrade it to a linear regulated power supply.


Crosstalk improvement
Cut the traces where it is marked in red, and solder a wire where marked in blue

[Deleted]

pin 4 and 5 of the tube (filaments) are connected.
This will ensure the annoying crosstalk is present in this amp too. when unmodded
just CUT this connection between pin 4 and 5 and this WILL cure the crosstalk just as oohms suggested in the picture above.
Should you want the BLUE LED out... just short it !

Converting the 6922 (ECC88) version to a 12AU7 version is possible.
I might make a tutorial for it in the future.
It involves cutting copper traces and adding some wires on the 6922 board too.
Bye, bye crosstalk !

I have Ian's amp running straight up to 50kHz instead of 10kHz (-0.5 dB)
just by popping in IRL530 (not IRF510, the 530 has half the Vgs voltage so more headroom)
and also replacing the LM317 for the higher grade LM317A.
The IRF510 will do the trick too but the IRL530 will increase headroom/output power also and clean up the highs roll-off in a similar way. ;D

Crosstalk can be cured (and distortion WAY lowered) by NOT using the filaments in the output path (you will need to supply them with their own voltage) and connecting the LM317's directly to ground.
The LM317's WILL need a MUCH bigger cooling fin and/or bias current of the output stage reduced in this case (do not take it below 70mA and you have to increase the output resistor from 47 to 68 Ohms). ;D
These mods will be used in Ian's amp.
Autobias works like a charm but is difficult to build even for more experienced D.I.Y.ers, so a simple but very effective mod it will be.

Be careful de-soldering the Bravo board.
The quality of the PCB is very crappy.
The copper traces fall off the PCB already when you just point at them.

replace the 25 volt power cap for a 35 volts type (will be MUCH larger)

the input decoupling caps are highly recommended when you have a DC coupled music source.
I would use a bipolar type and not polarised electrolite caps as these need a DC bias to work/form properly which is not present in this amp.
These caps will NOT be performing well unbiassed.

Also replace the input pot for a log type when you want better travel.

Mike's advice to swap the output caps for a better 105o type is also very adviseable (it seems only the resistors are still in place)
25 Volts will do in this case as the nominal voltage will be around 10-12 Volts DC

you can also replace the 47 Ohm output resistor for 68 Ohms.

[PinkFloyd]

Awesome stuff guys, many thanks.

Mike.

[Deleted]

Great Stuff ooms! Thank you!

I have two of these Bravo amps with the 12AU7. I just received a couple of Telefunken 12AU7/ECC82 tubes that, even without mods made this little unit sing! I listen to a lot of classical music, and the Telefunkens are simply fantastic with an incredibly large sound stage and a superb frequency response! They blow me away! Of the ten odd tubes that I have, these two seem to be the best!

I have the IRF510's on order and should get them in a few days. I have also ordered some Wima .22uF film caps to replace the ceramic ones in order to reduce the annoying popping noise.

I do already sense a problem with one of the units, and it is in the power supply end, which keeps cutting in and out. I cleaned out the input jack and the the plug, which improved it some, but I am very suspicious of the poor quality stock switch, and would like to replace it, but don't know what to order.

I see in the photos that you have replaced and relocated the original cheap switch with a sturdier toggle switch. Can you tell me the type and brand of the one you have used, so I can order it? Also, I would appreciate some step by step guidance, since I am a newbie, but with quite a bit of soldering experience.

Thank you again for the excellent post, and I look forward to more. What a great forum this is!

[oohms]

The switch is easy to replace. Any switch with 2 connectors will do, and if you have a 3 connector one, use the middle connector and one of the other 2 (doesn't matter which one)

www.jaycar.com.au/productView.asp?ID=ST0335&CATID=28&form=CAT&SUBCATID=448

But DON'T leave your headphones plugged in when you turn on or off this amp. The output coupling capacitor will charge/discharge through your headphones and there will be several volts of DC across them during this time. (Which can kill them, and explains the 'pop' heard when turning it on/off)

Best to wait 20 seconds or so between turning it on, and plugging in the headphones. And disconnect them just before you turn it off

[Deleted]

Thanks for the suggestions, oohms! I usually try to never use the switch with the headphones plugged in! I especially would hate to kill the HD600's which sound so good with this amp. Also, I always turn down the volume to '0' before plugging in, or unplugging the phones to protect my ears as well, for they can't be replaced!

[MaN227]

thx guys great info

and damn the black pcb is sexy as hell, as I look at my fugly ass green pcbs of the indeeds I think the black is head and shoulders above green. red would be awesome too , or even a nice blue like the hifidiy stuff has. green just looks cheap.

[Deleted]

Apr 8, 2010 5:58:04 GMT MaN227 said:

thx guys great info

and damn the black pcb is sexy as hell, as I look at my fugly ass green pcbs of the indeeds I think the black is head and shoulders above green. red would be awesome too , or even a nice blue like the hifidiy stuff has. green just looks cheap.

Yes, it's strange how colours affect our perception isn't it? I thought that when I saw it too! Black just looks plain slinky.

Ian

[Deleted]

the black Bravo board may LOOK nice (being black) but the quality of it is WAY less then the green PCB.
When desoldering the parts the islands often just fall off the PCB material.
It is the cheapest kind of PCB with the thinnest copper traces (cheapest)

If you do not plan to solder on it the black one will do.

[MaN227]

hehe man, I was just talking COLOR not quality. let me take an example of what I mean , look at PC mobo's and it'll be a cheap board if its green and black, red, or blue if its good. or should I say higher dollar board

bear in mind I'm someone who likes the purdy lights

[Deleted]

For those that bought this 12AU7 version and like to have the capability of bias adjustment for these amps too here is a 'simple' fix.

replace R13 and R14 (3k fixed resistors) to 5k adjustment pots.
use the centre tap and one of the other legs.
If you want to do it even nicer you can put a 1k fixed resistor in series with this pot.

DO start out with the pot in the middle setting before powering up !!

[Deleted]

If you want to do it even nicer you can put a 1k fixed resistor in series with this pot.

Even nicer could be a 2Kohm trimpot and a series 1Kohm resistor ?
This would give a finer adjustment.

[Deleted]

the range would be very small though given the fact that you could do max 3k resistance (which it is now) and min 1k.

the 5K would give more room for adjustments.

[Deleted]

Apr 11, 2010 11:25:44 GMT @solderdude said:

the range would be very small though given the fact that you could do max 3k resistance (which it is now) and min 1k.

the 5K would give more room for adjustments.

Frans
As the Anode current with 3K is only 0.4mA, perhaps the smaller range would be more worthwhile ?
1.There is a better chance of achieving a closer to optimum setting, despite the twiddler's best efforts.
2. Your original suggestion isn't idiot proof !


Alex

[Deleted]

5k pot with 1k fixed resistor gives a good range ! of 0.2 mA to 1.3 mA
your suggestion would give a 0.4mA to 1.3mA range.

the 0.2mA range is what might be needed for a wider range of tubes to be applicable (5k is needed to achieve that).

the ones that are doing this mod are not idiots so it does not have to be idiot proof.

if you leave out the series resistor the anode voltage in the worst case scenario (pot at 0 Ohms) would be around 23.5 volts and there would be no sound.
Just don't leave it there to long as the LM317 would be heated to quite hot temperatures before it sets itself to a lower value with it's internal thermal protection.
the anode current can't go any higher than that of the tube (max 1mA ?) so it really doesn't have to be that idiot proof.
Nothing can be blown up (the base current of the transistor is not going to exceed 10mA if you use a single 5k pot instead of the 5k pot + 1k resistor).
So in essence the single 5k resistor IS idiot proof in this case. ;D

like I said... use a 5k resistor and set it in the middle position before switching it on..
adjust the bias voltage after warming up for several minutes to somewhere around 13 to 14 volts.
Wait a few minutes and do the final re-adjustment and leave it as it is and nothing will go wrong.

You will find a 5k resistor will give EXCELLENT results and you can fine tune the anode voltage very well.

This is NO obligatory mod... only for those that want/need bias setting capabilities on the 12AU7 version and like to do some tube rolling as well.

[Deleted]

Great Stuff, Frans! Thank you!

Just what I was looking for. I have two 12AU7 Bravo amps, and would like to mod one of them to start with. If possible, I would much appreciate seeing some photos of the mods you suggest. Also, do you have a good source for the 5K trim-pots that can physically fit on the PCB?

I will be replacing the MOSFET's as well with the IRF510's which I already have, as well as the ceramic small caps with .22uF Wima caps.

I am a newbie here, but am learning a lot from this great forum, and it looks like everybody here is more than willing to offer helpful suggestions.

Thank you all!

P.S. I also have a couple of Linear Technology LT317A's that I plan to replace the stock ones with. I assume that it may be best to first do the trim-pot mod, and then do the rest, because the bias voltage most likely will change with the new 317A's in place, and will need to be adjusted.

[Deleted]

I suggest you do the oohms mods first (cutting of the copper traces and routing a wire as described in the first post of this thread)
IF you are able to get your hands on the IRL530 this would be preferable over IRF510.
Make sure it is the IRL530 and NOT IRF530 - IRF530N.
Same sound (more highs) but less distortion at higher levels as the voltage across the LM317 becomes 2 volts higher and won't come near it's 'out of linear operation current source point' as fast as you would with the IRF630 or 510.
The LM317's will be running somewhat hotter though as more voltage falls over them with the same current.

If you like the switch-on 'pop' a lot less 'intense and loud' there is an easy fix for that as well.
You can leave your HP plugged in when switching on in this case.
this involves changing the LED resistor from 1.5k into 10k AND mounting a 100uf/10V capacitor in parallel to the RED LED (mind the + and - connection)

< edit > pic's and tutorial posted 3 posts down.

I get all my electronics supplies from Farnell but there are cheaper sources out there.
Just look for something that can be fitted (7.62 mm, 3 mill)

[Deleted]

Thank you again, Frans!

I presume the R13 & R14 are installed towards the back of the amp, i.e. R13 is the one right next to the ceramic cap behind the power supply jack, and R14 is located behind the switch. Please correct me if I am wrong, or better yet, maybe as you suggested, perhaps you will be so kind and mark the locations in oohms's photos.

All the Best!
imagemaker18

[Deleted]

Frans
So further reducing Anode current isn't going to cause even more HF rolloff with the original FETs ?
Also, and quoting from JLH :
"An input DC bias voltage will normally be chosen so that an appropiate level of cathode current will flow and will maintain the DC operating point of the valve at a suitable position on it's Ia/Vg curve." (straight part)
This value is with typical twin triodes more like 2mA , NOT 200uA, hence why I do not believe we should further reduce the already very low Anode current of approximately 0.4mA.
Alex

[Deleted]

there will be more roll-off with standard FET's IF the anode needs MUCH re-adjustment when a lower current is used.
with IRL 510/530 this is NOT a problem though.
the anode current will very likely be around 0.4 mA at the optimal anode voltage anyway.

The tubes are used FAR from their ideal usage as I have already mentioned numerous times before, the anode voltage simply is too low..
Peeps seems to love this sound... let them..

the 5k pots are to adjust the anode voltage to it's proper setting, because there are NO cathode resistors this simply IS the easiest and best way to do it.
it's setting value will very likely be around 3k.
the proper anode VOLTAGE is needed to bias the LM317's (distortion) and for symmetrical clipping.
With a 5k pot you will have a wider range of tubes when rolling.

typical 13.5 Volts is near ideal (between 13V and 14 volts)

LM317A's should NOT have other currents then LM317.
it's just a higher grade with better tolerances.

HOW TO:

PICCA's from oohms (hope he doesn't mind me using them)



yellow ovals = 3k resistors:
replace with 5k pot or 2k pot+1k5 resistor (if you want more travel)

purple oval = 1k5 fixed resistor if you want a nicer 'softstart' replace with 10k resistor (LED's will light less bright !) and solder a 100uF/10V electrolytic capacitor over the RED LED as shown (mind the + and -)



hope this helps

[Deleted]

Thanks again, Frans! This helps a lot. Great suggestions!

[oohms]

I changed the mosfet gate resistor from 220R to 120R, and the output resistor from 47R to 68R, but the bass is almost completely gone. The top end does sound better, but there is way too little bass

Would it be the gate resistor or output resistor that contributes to this?

[Deleted]

Apr 17, 2010 9:53:23 GMT oohms said:

I changed the mosfet gate resistor from 220R to 120R, and the output resistor from 47R to 68R, but the bass is almost completely gone. The top end does sound better, but there is way too little bass

Would it be the gate resistor or output resistor that contributes to this?

Is the bass almost gone, or does it just seem that way now that the top end is more prominent ? I gather that you are more than capable of feeding tones in from a test CD or signal generator, which would confirm most likely that the bass has not been really reduced ?

[oohms]

I haven't gone to that level yet, but I'm comparing it to my SCHA, which i didn't think was overly bass heavy (using LM4562), and to what it was before i changed those resistors... I'll keep experimenting with different resistors and see what I can do (and hook it up to my new oscilloscope and have a look)

Another thing i have noticed, and its very subtle, is that when a 'punchy' sound is played, there is a slight distortion echo following it.

[Deleted]

It is always wise to do one mod at a time.
This way you'll know exactly what changed what.
Just put back the 47 Ohm (or solder 150 Ohms parallel to the 68 to make it 47 again) and have a good listen.
increasing the output resistance is more likely to give more bass then less bass.
changing the output resistance is likely to give the most sonic changes though.

the impedance of the anode circuit is around 30kOhm.. 220 or 120 ohms in series with that is not likely to give that much influence.. there might be other effects in play here ..

Neither mod is likely to cause bass to disappear..
that would more likely be caused by the input caps or output caps ?

lower the output resistance to say ... 34 Ohm (solder a 68 Ohm parallel to the present 68 Ohm output resistor) and see if that changes the distorting echo.
With a lower output resistance the driver is damped better.
Some HP's like to be driven at 120 Ohms some lower.
68 Ohms is a nice universal in between value.