Bravo 12AU7 Mods and Appreciation thread

[Deleted]

oohms wrote: 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.

Hello, oohms:

I was planning to install an inductor as you so suggest in order to also protect the 24V/2A adapter as well, but I was told that installing an inductor in series with the supply voltage may degrade the performance of the unit, especially in the high frequency range. Do you notice any deterioration in the HF region with the inductor installed as described?

Thanks!

[oohms]

To be honest i don't think there is a big difference, but theoretically its supposed to filter out the high frequency switching noise from the power supply

[Deleted]

May 8, 2010 5:49:52 GMT oohms said:

To be honest i don't think there is a big difference, but theoretically its supposed to filter out the high frequency switching noise from the power supply

Please tell me if I understand you correctly : there isn't a big difference in the HF switching noise, or there isn't a big difference in the total frequency response of the unit across the spectrum, i.e. while playing music?

Thanks for all your responses!

[Deleted]

The switching noise voltage from a SMPS (switch Mode Power Supply) is a so called common mode signal.
that means on the + and on the -output of the PS the noise signal is equally big and in phase and usually in the freq range from 30kHz to several MHz. and referenced to 'ground' (in short the noise voltage can be measured from the + to safety ground and from - to safety ground but NO signal (well actuall very little) between the + and - of the PS.
This becomes audible (sorry... may become) when the amp is connected to other equipment and it interferes with this equipment through ground (common).
In this case a single coil does not do much as the gnd plane is unfiltered.
For higher freq (above 10MHz) certain ferrite clamps can be of assistance.
For optimal use of these loop the + and - wire next to each other 3 times through a ferrite where the wires fit through snuggly.
More windings = less effect, less windings is less effet, wires loosely through the centre = less effect.

What you need to do in this case (If you hear whisteling noises when you connect it to source equipment otherwise it's pointless)
is to install a common mode filter.
Similar to the one oohms made for this amp BUT the wires from the - (gnd) of the powersupply need to go through also.
You can buy ready made common mode filters but because the DC current that flows through it you'll need one that does not saturate below 1A, otherwise it won't do a thing.
On the input and the output of the filter you connect 100nF filters across the power line.

If you don't have whisteling noises you don need to do it.

The filter WON'T affect HF properties of the amp.
What it may do is keep distortion lower and sound cleaner IF noise IS there.
Like oohms said ... theoretically it suppresses (not common mode) noise above say... 10kHz.
Most SMPS usually work at freq. between 30kHz and 70kHz and thus the ripple can be found there too (a few mV usually) but the Bravo amps are REALLY VERY susceptical to noise on the powerlines.

If you can... use them with a linear regulated power supply with a transformer in it.

Hope this isn't too technical.

[Deleted]

Hello, Frans:

Thanks for the elaborate explanation! It was a little technical, but from it I understood that the HF response of the amp will not be affected to the point of being audible.

This is perhaps slightly off-topic, but my experience so far has been with modding the Indeed 6922EH v2 and the Indeed G2 6N11 amps, and I am asking these questions in preparation to modding the 2 Bravo 12AU7 amps that I have. On the Indeed 6922EH v2 I replaced the two 1000uF filtering caps, connected in parallel to the 6800uF with two 2200uF, giving me a capacitance of 11200uF, replaced the two 1000uF output electrolytics with 2200uF, the stock MOSFET's with IRF530 as you suggested, and the LM317's with Linear Technology LT317A's, also the 100nf bypass film caps with 3.3uF Wima film caps. The sound improved tremendously with the Telefunken tube that I am using, but I found that if I would adjust the bias to 16.5V, the 317A's would reach a point where they would cut-off the supply and the amp would go dead. After lowering the bias voltage to 12.5V, all was well, and I did not notice any audible change in performance. I am a classical musician, and can modestly say that my ears are good, and well trained. I did similar mods to the Indeed G2, except I replaced the single 6800uF filtering cap with a 10000uF one, but in trying to drill the Plexiglas my drill slipped and shattered the top panel. I emailed Peter from Indeed Audio, and told him what happened, and while he kindly is sending me another top panel, he advised me not to use such a large electrolytic capacitor because, with the low impedance supply circuitry upon power-on, the initial voltage surge may damage the 24VDC/2A adapter. I then asked him if he thought that an inductor/coil in series with the supply voltage would help protect the adapter, for it will increase the impedance of the circuitry. He responded stating that it might degrade the performance of the amplifier, especially in the HF range. He suggested that the best way to do this would be to use a traditional transformer, however I already have at hand a few toroidal 320uH inductors, so I would like to try using them. ;D

Pfew!!! Sorry if it this has gotten too long! All of the above questions are in preparation to making similar mods to my 2 Bravos, using the 10000uF filtering caps (great bass response) , etc., so on second thought, this post may not be so far off-topic.

Kindly, let me know about your thoughts on this.

[Deleted]

I recommend IRL530 or IRL510 as these have the lowest Qgs and best HF performance.
another advantage of the IRL types is the LM317's remain longer in their linear operating area decreasing distortion because of their low Vgs.

High value caps are more likely to damage the on-off switch then the PS itself IMO.
These things are short circuit protected and start-up surge is only for a very short time.

The coils won't help much as the inductance you would need to soften (smear) the puls would be very high and so will the impedance be.
The big cap on the PCB is where the low impedance comes from and is not dependant on the PS so the coils will not affect HF.

I once experimented with special capacitors in the power supply.
These are specialised caps and are dedicated to be shorted and to deliver VERY fast transients.
You can find these caps in flashers for camera's.
their sole purpose is to deliver very much power in the shortest possible time (a flash can be 1/10,000th of a second).
VERY low ESR and MUCH lower inductance compared to normal caps.
These are around 47 to 100uF and 400V types.
Just use a 35V type cap (4700uF for instance) in the Bravo and put one like THESE in parallel.
Now THAT's transients and HF response for ya !!

Try making a flashgun with normal capacitors, even the best caps cannot supply the fast transients that these things can.
I am really amazed that I never read about this in any forum/mag/article.
I sure as hell can't be the only one experimenting in Audio, Photography and Electronics.

short a normal cap (with a screwdriver) and you here a 'poff' and maybe a 'paff' if you know what I mean.
Short one of these guys and you hear a sharp 'TAC' and a intense spark.
How's that for an idea ?
Give it a go ..

All my power amplifiers have these caps parallel to the PS for over 20 years now..
never tried it on a HP amp. I used only speakers in that period.
HP's is only a thing for me in the last 15 years or so.

replacing the output caps to 2200 is likely to give the most increase in bass response though.

[remolon]

A modification to preserve the life of the 12AU7 is to substitute R6. With the 7R5 resistor the filament current is 167 mA (1,25/7,5).

The recommended value is 150 mA and it is achieved with a 8R5 resistor.

[Deleted]

You are quite right for the 12AU7 version.
12AU7 = 300 mA (2 x 150mA) so a 8.66 Ohms resistor would be optimal. (2 x 145 mA + 13 mA LED current)
8.45 Ohms and 8.5 Ohms is acceptabel too.
The 7.5R resistor is for ECC88 which has 365mA glowing current. (2x 168mA + 13 mA LED current)

[Deleted]

Hi again, Frans:

And Thank you, remolon, for the mod suggestion!

I would appreciate it if either of you point out where exactly on the PCB is R6. I think I have been able to trace it via pins 4 & 5 which are the heater pins, but would feel better if one of you, experts guide me by perhaps using one of oohms' photos.

Thanks for all your helpful suggestions!

[Deleted]

the 2 resistors you need to change to give the tube a longer lifespan are in the black ovals.
purple-blue-black-silver rings = 7.5 Ohms
replace by 8.66 Ohms resistors.

[Deleted]

Frans
1 W resistors , or 2 x suitable parallel .5w or .6W (perhaps 2 x 18R .6W MF types) ?
Alex

[Deleted]

the same size (0.6W) MF (metal film) resistor (8.66 Ohm = standard 1% type from E48 and E96 range)
IF you want to approximate by using 2 E12 range resistors (most common) in parallel the closest thing will be 10 Ohms//68Ohms (one on top of the PCB and the other on the bottom ?)
This will give 8.7 Ohms.

the power dissipated by the resistor itself = 0.15Watts b.t.w.

[Deleted]

May 11, 2010 8:11:59 GMT @solderdude said:

the same size (0.6W) MF (metal film) resistor (8.66 Ohm = standard 1% type from E48 and E96 range)
IF you want to approximate by using 2 E12 range resistors (most common) in parallel the closest thing will be 10 Ohms//68Ohms (one on top of the PCB and the other on the bottom ?)
This will give 8.7 Ohms.

the power dissipated by the resistor itself = 0.15Watts b.t.w.

Thank you, Frans!

I will try to do the above, i.e. a 10 Ohm and a 68 Ohm in parallel, for it seems, the 8.66 Ohm resistors are harder to find.

[Deleted]

May 8, 2010 21:57:38 GMT @solderdude said:

I recommend IRL530 or IRL510 as these have the lowest Qgs and best HF performance.
another advantage of the IRL types is the LM317's remain longer in their linear operating area decreasing distortion because of their low Vgs.

Thanks for the suggestion, Frans! I finally took delivery on several IRL530's, and replaced the IRF530's on my Indeed G2 with the former. The difference in HF response is dramatic! I did an A-B listening test with my other Indeed, which still has the IRF530's installed, and, while on first listen, the latter sounds pretty good with seemingly flat response, ::)upon switching to the Indeed G2 with the IRL530's the whole spectrum extended, especially in the HF, creating the impression of a window opening up, letting the live performance of a 100-piece symphony orchestra in! This spurs me on to replace the IRF630's on my 2 other 12AU7 Bravo amps, as well as the IRF530's on my other Indeed.
Great stuff!

[meins321]

Hello i have got a Bravo G1 too and wanted to ask what types of capacitors everyone uses for the 1000uF and 6800uF as they are a tight fit.

mine is still stock, only the 1000uF are new Panasonic FM Low ESR as the old ones fell appart into pieces ;D

Second thing would be, where to buy a LM317A in Germany i can only source LM317 without that special "A"-Grade and is it important?
And for the Mosfet IRF510 is the only thing i could get, somehow the 530 is always the wrong one with "N"

What size should i use as "gate" resistor, i read though the thread an i looks like some people lost their mid's and bass response that would be pointless then to replace the IRF630s

[Deleted]

The LM317 A has tighter specs.
I doubt they sound different..

You should try to find IRL510 instead of IRF510
the IRF530N doesn't do a better job the IRF630
When using IRL the LM317's get a lot hotter, but the crosstalk and distortion will be much better.

6800uF/35 will be much bigger and won't fit.
For a mechanical fitting 35V cap you will have to use 3300/35V (perhaps 4700/35V)
If you want bigger caps in there you will have to modify the topplate or fit them elsewhere.
KEEP the wires VERY short !

leave the gate resistor at 220 Ohm not much to gain there.

[meins321]

oh i just forgot, my blue led was dead from start and as i read on the other thread it could be shorted or desolderd..... can i leave it in place as the amp is working without it from start? (never noticed that as the blue led was a no-no feature for me)

and i will add 2 Nichicon 3300uF 35V 105°C in Parallel to the back and reposition the powerplug (removing the stupid switch and replacing it too as it is getting rusty lol ).
+
my 1000uF 35V 105°C Panasonic FM capacitors should be fine then :-P
(they are dxH= 17mmx25mm)

EDIT: the Mosfet i could source
www.conrad.de/ce/de/product/162789/TRANSISTOR-HEXFET-IRL510-TO-220-IR
www.conrad.de/ce/de/product/162795/TRANSISTOR-HEXFET-IRL530N-TO-220-IR
www.conrad.de/ce/de/product/162798/TRANSISTOR-HEXFET-IRL540N-TO-220-IR
look and compare the last two, i thing the last one might be useful ( wrong description?) as the RDSon is very low ^^

i think i got a trimmed down version
as it is rusty on the phoneplug and switch after a short time. ( And no i didn't water it myself, it is green patina thus ni that was used instead of metal and silver on it )
+
no ceramic caps at all on my pcb ( wima here we gooo!)

[EDIT2]
Whats the difference between IRL510 and IRF510?
I have some good/new IRF510 here but could buy IRL510 if its a ultra big difference ^^

[Deleted]

You can leave the Blue LED as it is or short it (preferrred) as it is probably shorted itself already.

RDs on is of no importance in these amps as it is only important in switching applications with high currents.
It's the Qgs that needs to be as low as possible.
Qgs is the charge that is needed for the gs(Gate Source) capacitance to charge so that the S can 'follow' the G.
This GS capacitance is charged by the current flowing through the tube (anode current).
This is a very low current (because of the low voltage it operates on) of 400uA.
At higher frequencies (10kHz and above) this current is not able to fully charge the GS capacitance (of the original IRF630) and thus the S (=output of the amp) cannot follow the gate anymore resulting in high frequency roll off.
You cannot change the tube current so the Qgs needs to be as low as possible.
the resistance of the gate resistor (220 Ohm) is so low compared to the equivalent resistance of the Anode circuit (about 30kOhm) that it does NOT influence the HF roll off even if you half or double it's value.
IRL has lower Qgs AND lower Vgs (half that of IRF).
because of the smaller voltage drop of the IRL (with the same anode voltage, which ideally should be between 13 and 13.5 volts)
the LM317 remains longer in it's operating point (acting as a true current source) and thus the distortion won't pop up until you play really loud.

You can use the IRF510 too but the IRL has more benefits.
IRL keeps the bad distortion (non linear) further away and extends the freq. range slightly further then IRF

because there is a higher voltage (IRL) across the LM317 it will get MUCH hotter.
I suggest to change the cooling fin to something bigger that will still fit (preferably black, this cools slightly better)

the www.conrad.de/ce/de/product/162789/TRANSISTOR-HEXFET-IRL510-TO-220-IR is good.
Caps are good too.

The power switch needs to be replaced as it breaks anyway within a few months.
Very crappy mechanical and electrical properties not suited to accomodate the high switch on currents because of the huge capacitors used.

[meins321]

^^ always good to have a trash box for parts.
got some of these: search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=IRL530PBF-ND

as their (Qg) @ Vgs is 28nC @ 5V remains the same it might they way to go IRL530PBF-ND

[Deleted]

IRL510 GS charge = 2.6nC
IRL510 total Gate Charge = 6.1nC
IRL510 Vgs = 2V

IRF510 GS charge = 2.8nC
IRF510 total Gate Charge = 8.3nC
IRF510 Vgs = 4V

IRL530 GS charge = 3.8nC
IRL530 total Gate charge = 28 nC
IRL530 Vgs = 2V

IRF530 GS charge = 4.8nC
IRF530 total Gate charge = 34 nC
IRF530 Vgs = 4V

IRF630 GS charge = 7.5nC
IRF630 total Gate Charge = 45nC
IRF630 Vgs = 4V

IRL510 is the fastest one (this one gives a very nice -0.5dB @60khz) very well extended
IRF510 is next in line but has double the Vgs. (this one gives a nice -0.5dB @55khz) well extended
IRL530 has half the Vgs as IRF. (this one gives a very pleasant and full sounding -0.5dB @30khz)
IRF630 is the worst ! (this one gives -0.5dB @10khz) warm but rolled off signature

There are plenty more possible replacements.
There is differences in manufacturers with the same typenumber !
These specs are from IR (International Rectifer)

IRL keeps the nasty distortion further away then IRF

[shattermyillusion]

Hey solderdude, you seem to know alot about the different mosfet's that you can use to modify the Bravo V2. I have a question, my local store stocks Fairchild mosfets, more specifically the FQP13N10. Supposedly this cross references to the IRL530PBF. Would this give the desired improvement that the IRL530's do in the Bravo V2? Here is a link to the datasheet: www.fairchildsemi.com/ds/FQ/FQP13N10.pdf

[Deleted]

shattermyillusion
Frans is taking a well earned break for a couple of weeks, so he may not reply for a while.
SandyK

[shattermyillusion]

Okay cool, thanks for letting me know. I also just found out they have limited quantities of IRL530a's...again from Fairchild Semi. Not exactly sure how close this is to the International Rectifier version. =/

[shattermyillusion]

Well, I went ahead and swapped in Fairchild IRL530a 's into my Bravo V2. I noticed when I pulled out the original mosfets they were IR IRF630N's! Anyway, I'm no audiophile, and this is my first amplifier, but as far as I know, it does seem as if the highs are more crisp and detailed with the new IRL530a mosfets.

Spec wise they seem similar to the original IRL530 proposed:

Fairchild IRL530A
Qgs: 2.7nC
Qg: 24nC
Vgs: 2v

IR IRL530
Qgs: 3.8nC
Qg: 28nC
Vgs: 2v

[Deleted]

Fairchild IRL530A is excellent !

FQP13N10 has 4V GS voltage so the IRL is a better choice, FQP13N10 is more like the IRF530.

I recommend the crosstalk mod (page 1) and replacing the power cap for 35V type and replacing the power switch (crappy quality bound to break anyway)
The output caps can be replaced for 105oC types.
Also if you want the bias to be adjustable replace the 3k resistors for 5k pots. (see thread)