Reformed Jadis 300B SE monobloc amps Nov 2012.
A customer recently purchased a pair of second hand Jadis SE300B amplifiers officially rated for 10Watts.
Here is a picture of what he purchased for $6,000 in September 2012.


These mono amps were made in 2004 and the picture shows them without their two
metal cages with Allen bolts to fix them over the 2 rectifier tubes at rear and
2 x 300B + 6SN7 at front.

The description from Stereophile included this :-

Tube complement (each): 2 x 300B, 2 x 5R4, 1 x 6SN7.
Output impedance: adjustable, for 1–16 ohms.
Output power at 1kHz for 0.3% THD: 10W rms (10dBW).
Power bandwidth at 8W: 40Hz–15kHz &#150 dB; 40Hz–30kHz –3dB.
Input sensitivity for rated output: 450mV rms. Input impedance >100k ohms.
Power consumption 140W each.

Dimensions: 8" (203mm) W by 9" (229mm) H by 20.5" (521mm) D. Weight: 65 lbs (29.5kg).
Price: $13,000/pair (1996); no longer available (2012).
Manufacturer: Jadis S.A.R.L., Villedubert, France.

Unfortunately, the information above is rather misleading for the two amps which
were brought to me after purchase by the owner who said there was bad hum in one channel.
He wished that I fully examine them to ensure they would be reliable and provide fine hi-fi.

Problems were :-
(1) I found the hum on one amp was due to the 0V rail being directly connected to chassis. 
When the 0V rail was re-connected to chassis through a 22 ohm R, hum was much reduced.
The other amp didn't make hum, but both were given 22 ohms which Jadis should have used.
Then I found many other design shortcomings and very serious design mistakes and quite
unacceptable quality control for the OPT and the B+ filter chokes.

(2) Just one common 5Vdc x 2.4 amp DC supply for both 300B.
(3) Just one common R&C cathode bias network for both 300B with Ck = 4 parallel x 47uF/450V,
(OK) and Rk = 300 ohms (Not OK), although 300r = 9 x 2k7 w/w vitreous enamel, each 4W.

(4) Two silicon diodes were soldered to 5U4 B+ rectifier tube socket for B+ supply to the pair of 300B.
These diodes were shunting the tube diodes, so this 5U4 was doing nothing except produce wasted heat.
The 5U4 is rated for use with C1 = 40uF and if HT Va-a = 600Vac, then expect +335Vdc at 150mAdc.
The data suggest that if Va-a = 670Vac, which is the Jadis HT value, you'd get +374Vdc at 150mAdc.
But Jadis have 470uF + low value choke + 470uF as the CLC, and Jadis had 5U4 charging 22uF then
10r before the 470uF etc, so tube diodes would have not lasted long due to excessive peak charge
currents, so I can see why Jadis must have abandoned the idea of having a tube rectifier for the 300B
supply.  With a tube rectifier, Ek = approx +55V, Ea = approx 310V, B+ = 365V, Ia 183mA.
Because Si diodes had been used, B+ was raised to +430Vdc regardless of whatever increase in Ia.
With Rk = 300 ohms,and Ra for both 300B = 350r, and with Rk at 300r, effective total Ra at dc =
1,550r, so increasing B+ by +65Vdc increases Ia by 42mA to 225mA so Ek = 67.5V, Ea = 362V, so
each tube has Pda = 40.7 Watts. So just adding the Si diodes without the increase of Rk would soon
cook the 300Bs to death, and of course they don't stay matched, so one runs hotter than the other,
and soon they fail from too many cycles of running too hot. While the Rk of 300 ohms would be OK
in theory with a tube rectifier, its no good in the Jadis. In the one amp which did work a bit, the Ia
difference was 50% and one 300B was far too hot, one was cool.
0.16 Amp fuses were fitted between each 300B anode and the OPT primary. With Ia at 160mA,
and Ea at 300V, Pda = 48Watts before a fuse blows so 300B can become damaged before the fuse blows.
It seemed to me the soldering style was the same at the diodes as for the rest of the point to point wiring,
so I suspect it was done by Jadis and not the previous owner or by other ignorant idiot.
The other amp kept blowing anode fuses after turn on. The 300B had become already damaged internally
by excessive grid heating which makes the grid wires become deformed. 300B have grid wires arranged as a
rectangular helical tube, and wires have a straight length which tends to become bowed after grids have been
overheated. Eventually, such deformation causes arcs between anode and grid or between cathode and grid.
300B are nice tubes, but are very fragile compared to using a KT88 strapped as a triode.

(5) The 6SN7 has two triodes, one for input stage and other for driver stage and both stages had identical set up
with RLdc = 150k, and fed from a +400Vdc B+ supply rail from a second HT winding and 5U4 rectifier.
No silicon diodes were used to boost the B+. Ia in each 1/2 6SN7 was about 1.5mA only, rather too low.
The 300B require up to about 55Vrms of drive voltage at their grids. The 6SN7 single driver triode struggles
to make this amount of drive voltage with low THD.

(6) No global NFB was used so that output resistance is too high. Anode loading was too low so THD was high
at full power - if full power could be obtained - if the 2 x 300B each had equal Ia. As the amps were, the one which
barely gave 10W and THD was too high. 

(7) The anode load for each 300B for maximum PO would have been about 2k0.
With 2 tubes, primary load total = 1k0, including OPT winding resistance of approx 5%.
With OPT set for "4 ohms" and a 4 ohm load, OPT TR = 14.9:1, so ZR = 222:1 so RLa = 888r + 44r for Rw
= 930r, a trifle LOW.
So really, there is not a good match for 4 ohms because most 4 ohm LS have minimum Z of maybe 2.5r
within the main power band. 

OPT has 4 secondary windings which may be re-arranged in series-parallel to give a range of turn ratios
of 29.8:1, 14.9:1, 9.9:1. and 7.45:1.
These give load matching according to the following table :-

4 // N = N
Sec load = 1.0r
RLa = 932r0
2 // ( N +N ) = 2N
14.9 222
Sec load = 4.0r
RLa = 932r0
N + N + ( N // N ) = 3N
Sec load = 9.5r
RLa = 932r0
N + N + N + N = 4N
Sec load = 25.9r
RLa = 932r0

Now with the extra high Ea, the Ia should have been allowed to be lower by raising Rk
from 300r to about 500r at least and then the load values for full power need to be twice those
shown in the table.
I have used dual Rk of 1k0, giving Ea = 355V, Ia = 70mA, Pda = 25W, and tubes are comfortable.

The amp had OPT strappings set for 9.5r, and so speaker needed to be 16 ohms with raised Ea.
With amp set for 4.0r, amp needs to have 8.0r, and there isn't really a good 4 ohm match possible
if the original OPT is used.
With silicon diodes at rectifier tube the B+ is higher and it becomes obvious that it is so
very easy for these amps to be used by an ignorant stupid audiophile who doesn't have any idea
about load matching and he manages to overload and damage the tubes.
(8) The OPT in BOTH amps were found to have NO AIR GAP.
After modifying the circuit so that each 300B Eg1 bias voltage could be adjusted,
I was able to get both Ia equal and not too high and then measure the amp audio performance.
I could not believe what I was measuring with bass response because it was the worst I have
ever seen in any SE amp. So I tested the OPTs when connected up to an external test circuit
as follows :-

As you can read from the image info, the usable response was effectively between 220Hz and 25kHz.
Lp was measured at 50Hz to be 1Henry when it should have been about 20H. The core was quite saturated with
Idc at 160mA. With no Idc, Lp became 81H, so obviously there was no air gap, and laminations are fully interleaved
as for a PP OPT.

(9) The B+ rail for 300B had CRCLC filtering arrangement. C1 = 22uF, then R = 12 ohms, then C2 = 470uF.
L was < 50mH.
I removed the E&I laminated chokes which have low Rw = 15r, quite low, indicating very few turns. Core size
has tongue x stack = 20mm x 20mm approx. I pulled them apart to check the air gap. I found the gap = 1.5mm
and 10 times higher than what it should have been after a few calculations. I adjusted the gap down and
re-tested the L and increased it to 0.8Henry with Ia at 150mAdc. The choke could have had Rw = 30 ohms
using thinner and more turns which would have given L = 2H, but 0.8H was a vast improvement.
With Silicon diodes there was no need for C1 of 22uF, so I removed this, but replaced 12r with 18r x 10W
placed between HT CT and 0V, so that diode charge currents are limited and to prevent switching spikes
in 0V rail finding their way into audio paths. So filter became C1 L C2 with 100Hz Ripple voltage at C1 470uF
= 0.6Vrms and at C2
= 4mV, and quite acceptable for connection to the OPT primary.

(10) The huge oversize Jadis power transformer is twice the weight it needs to be. There are two HT windings
one for the +440Vdc rail for 300B, and the other for the +400Vdc rail for 6SN7. Both are adequately rated.
The lesser HT winding could supply perhaps 40mAdc easily if someone wanted to use an extra and separate
driver tube such as EL34 in triode so that the 6SN7 could become a stand alone paralleled input tube.
Such a better arrangement is what I would expect after paying $13,000.
Then there are 4 low voltage secondaries, 2 x 9.6Vac, and 2 x 5Vac, all rated for 4Amps.
1 x 9.6Vac is rectified to make 12Vdc and followed by series SS regulator for 5.0Vdc for both 300B.
The regulator was on a heatsink under the OPT in the chassis space, and it liberated about 12 Watts.
The other 9.6Vac is used for making a regulated 6.3Vdc for input 6SN7. 5Vac windings are for the
2 x 5U4 cathodes.
I decided each 9.6Vac windings would be used for individual separate cathode heating of 300B
so separate R&C bypass networks could eliminate the very poor biasing the amp has. Instead of 2 regulators
I used CRC with R on heatsink placed where 5u4 used to be. Also on the same much larger heatsink
positioned for much better ventilation and less heating effect on sub-chassis items, I placed the 2 x 1k0
Rk to each 300B. So, heat is now managed better.

(11)  I found the amp had 2 x 160mA anode fuses for the pair of 300B. They didn't blow until
damage was done in one tube and then the other would blow just after. Mains fuse would not blow
unless something much worse occurred. Blown fuses under the chassis were difficult for an owner
to replace, and 160mA is not a common value. I abolished these damn fuses!
I have installed active protection by monitoring cathode current with 22r between each R&C cathode bias
network and 0V, and have provided 2 test points on the side of the amps so that Vdc across 22r may be
measured without moving the amps. With Ia = 70mAdc, V 22r = 1.54Vdc. Its a safe measurement to make
and if anyone shorts the test point to 0V, it won't kill anything or anyone.


I decided to remove both 5U4 and the 2 sockets. I have replaced then with 8 x Si diodes under the chassis.
Using pairs of 1N5408 in series, Current rating allows 3A and PIV rating is 2,000V, so they are going to
last OK, and if a short occurs, a mains fuse will blow, and not the diodes.
Where the tube rectifiers were, I have mounted a heatsink for both 2 x 1k0 Rk for 2 x 300B and 2 x 5r
for CRC filters for two 5Vdc cathode heating supplies for 300Bs. The 2 x 5Vac windings have been seriesed
and the 10Vac rectified to make 6.3Vdc x 0.6Adc for 6SN7 heaters.

I removed both Jadis OPTs and replaced them with Hammond SEA1627. These gave Lp = 18H at low
level signals, and Z matches of 2k5 : 4, 8 and 16 ohms. Using the 8 ohm tap, and using load = 5.3 ohms,
PO max was nearly 16 Watts, so that at least 10Watts was available for any load between 3 and 11 ohms.
This was a far better outcome than the original Jadis OPT. BW was much wider. I removed bell ends from
Hammond OPTs then fitted AL angles for mounting with new pots the same size as the pots around
Jadis OPTs. I used a 50:50 potting mix of casting resin + sand to make a concrete around OPTs.
The Jadis badges will be transferred. So the amps will still look fine. There will be a note
glued to OPT about what's inside. Load match changes will probably never have to be made, but now
it means just moving ONE WIRE from the output terminal to one of 3 available taps while using a soldering
iron. The original change of load match required
altering pattern of strappings held with 8 x 4mm nuts
on bolts. 2 new straps needed to be made.
The anode load with sec load = 5.3 ohms is 1,656 ohms plus winding resistance estimated at 132 ohms
giving total anode load of 1,788 ohms, so each 300B has a theoretical load of 3,500 ohms, nearly twice the
original design load so damping factor is much better and THD much reduced.

Only 2 of the original 4 x EH300B remained usable during many tests. The other two destroyed themselves.
I fitted 22r current sensing R to check Idc and Iac in the 2 usable 300B with individual Rk biasing.
The difference in Iadc and Iac was less than 10%, far better than when tubes shared the same Rk.
The difference in Iac means that one tube produces different load current than the other for the same Va
so the loads each tube is effectively working with is different. As long as the RLa for each does not
vary by more than 10%, all is well. The condition of grid and emission of cathodes affect the load power
of each tube.

The two bad 300B degraded during testing and both would arc between grid and anode. Eg1 would rise to
+330Vdc, Ik would then rise, and rise, and eventually an anode fuse would blow in one tube then the other.
I watched and turned off the amp. With active protection fitted the turn off becomes automatic. 
These 2 tubes thus became rubbish.

But at least 2 x EH300B did work OK well enough and music was quite splendid after
totally gutting the amps and re-wiring everything as the new schematic shows :- 

I hope the above schematic shows enough details to allow anyone to build a good 16W SET amp
with a pair of 300B. The unusual feature above is the use of a constant current source for V1B
driver triode's anode supply. If a DC carrying R was used for 5.1mAdc, the value would have been
34kohms. But then the load offered by the two following parallel 150k grid bias R17, R18 would have made
total DC and CR coupled load = 26kohms, and this a value I consider too low to get my wanted
best linearity and wide voltage swing using just 1/2 of a 6SN7, and while keeping Iadc at a healthy
5mA, and have Ea high enough to get the swing of 70pk volts. So, with a CCS DC supply,
the load is only 75k for R17, R18, and then the distortion is much reduced and the single triode
just works a lot better. The input 1/2 triode, V1A, did have Rdc to anode of 150k, same as the
V1B. I thought the Idc of less than 2mA was too low for both the identical stages for input and driver.
But the input need only generate a small voltage, and so a moderate rise in Iadc was called for,
without reducing the total anode load ohms too much, so you see my result of thought.
The sound from any any amp like this dependent on the sum of the integrity of all aspects
of design, including finer points of the way input ans driver stages are set up.

The power supply shown uses the original Jadis power tranny which looks suspiciously like
it may have been meant for some other amp requiring more power. Solid state regulators were
used to make the Vdc needed for heaters. But when I decided to use TWO separate 5Vdc supplies
so that EACH 300B had its own 5Vdc supply, I went to the simpler arrangement shown with
CRC filtering. There is less to go wrong, and I needed one extra Vdc supply than in the original
amp. This meant using the 2 x 5Vac windings for heating 5U4 cathodes as a Vdc supply to the
heater in 6SN7. It occurred to me, that if someone were to replace the 2 x 300B with say
3 x EL34 in triode, then the 2 x 9.6Vac windings could be used in series to make 19.2Vac,
and this means you'd get 6.4Vac for each EL34 filament in series! I bet Jadis hadn't thought
of this, but it is possible, if one likes changing the chassis to suit 3 x EL34, or better, 3 x 6CA7.
Load line analysis tells me about 18W is possible, with same Hammond OPT and the same B+.

The protection circuit above stops Ek rising too much to damage the 300B. There are those who think a
300B is a fine old rugged tube design; It damn well ain't rugged !!!! They break real easy, and if the tube
is subject to severe grid input signal, the grid can overheat and the fragile box-section grid wires can
become bent in heat and arcing can occur between anode and grid. This happened to 2 x EH300B
during my tests. The original amp had just one 300 ohm Rk for both 300B, and Ia was unbalanced and
way too high for at least one 300B. The standard type of 300B such as EH does not like being used
wrongly with Pda above Watts, and 28Watts is my limit.
The owner has replaced EH300B with the smaller version Emission Labs 300B; they make different
versions of 300B.
The smaller version looks like a well made rugged 300B. Its cathode takes much longer to warm up than
for EH, and seems less microphonic. EH300b, like so many others, dings like a bell when you tap it likely
because the metal grid etc vibrate easily. The EML seem far more rugged. The EML cathode has lower
resistance when cold, but when hot, the cathode has almost the same R as EH. Don't ask me what EML
have done with metal properties, but I prefer a slow starting cathode.

The schematic shows 1.54Vdc is normal at R7, R9, and if it rises to just over 2 Vdc, about 0.6Vdc appears
at SCR gate which turns on the SCR to power the relay which interrupts the mains input winding.
In this schematic, unlike others before now, I have used a bi-color green-red LED pair in the one 5mm
dia package to replace the original 5mm green LED in an holder at the chassis front plate. I doubt
I could make the circuit simpler. This circuit could be used for an amp with any number of output
tubes because the SCR gate reacts first to the highest Vdc signal from any cathode. If only 1 tube of a
dozen becomes faulty with too much Ia, the amp turns off. The fault detection is simple, and does not
include any ability to detect a tube which has Ia too low. And the circuit does not tell an owner which
OP tube has a problem. However, I have mounted 2 test points, P & Q, to allow monitoring the
Vdc using an cheap voltmeter on its Vdc range. Some owners just will not cope with such a simple
easy technical operation. But condition of 300B can be checked now without moving amp, turning
it upside down on carpet, removing bottom cover, then risking death from electrocution while
probing around. The two 160mA fuses in each 300B anode circuit have been removed - no need
for them any more, and so now all risk of shock has been removed.

New circuit boards and terminal strips have been glued to chassis with Selleys 401 Silicone, rated for 200C.
Not bad stuff. Screws hold some boards to stainless steel as well. Boards are from 6mm hardwood marine ply
and strips are 12mm x 10mm hardwood. I have used 1mm copper wire tracks for protect board, hooked under board.
Other terminals are 4g x 16mm long brass plated c/s wood screws used for small cupboard door hinges,
usually easily available. The soldering iron heat cauterizes the timber to stop it splitting over time. The effect
is that the screws are every bit as good or better than fancy turrets in PCB boards. All timber has a generous
coating of polyurethane varnish applied. Come back in 70 years, and the circuit should still be serviceable.
Everything under the chassis now operates a lot cooler because hot running SS regulators are removed,
and replaced with generously rated resistors glued to heatsink above chassis.


Here is the completely new wiring under the chassis. It is more complex than the original chassis parts,
but allows better access to parts. Wires are lashed up tidily to stop it all looking like a rat's nest.
GONE is the extensive use by Jadis of flying leads on stand offs and other cheap nasty ways of
wiring and soldering. 

This shows the placard placed to indicate terminals for load match changing. There is only ONE WIRE which
needs to be moved to suit the speaker to be used. Its now set for 6 ohms. But if someone had a speaker of
4 ohms with dip in band to 2 ohms, the 3 ohm terminal is best. But a speaker with nominal Z of 4 ohms or above,
even 16 ohms, may be used at 3 or 6 ohm terminals. If 16 ohms is used on 3r, power is limited to 5 watts,
but if speakers are sensitive like old Tannoy or are horn loaded, the the 3r terminal should offer high enough
levels at the highest damping factor and minimum distortion.  Most people will never need to change from the
6r terminal. The amp could have had 4 rear binding posts, Com, 3r, 6r, 12r, but I could not easily source
bind posts matching those used by Jadis, so hence the adjustment is made under-chassis.  It thus is less
confusing for any owner. Owners have a terrible habit of always using a 4 ohm speaker plugged in between Com
and 12r, and thus damaging music and their tubes. They only learn through pain, and still never know anything
about ohms, resistance, voltage, current, or Ohm's Law. Better that I minimize the likelihood of an owner
making a bad mistake. If a mistake can be made, you can trust an audiophile to make it!  


The above shows the heatsink above chassis for cathode biasing resistors and the R used in CRC filters for
300B cathode heater supply. The original metal mesh cover box was later fixed back over this heatsink.
Appearance is slightly changed with no 5U4, but it looks OK.
Black capacitors are rather nice quality 470uF x 450Vdc rated, well selected by Jadis.


Here we see Emission Labs 300B plus NOS RCA 6SN7. Jadis had 4mm dome head nuts on bolts
securing sunken ceramic tube sockets. But EML 300B have a rounded shoulder on glass
at bottom of tube near tube base, and this prevented tubes being fully inserted to sockets.
Dome nuts had to replaced with plain nuts with less height, and no more fouling the glass on
tubes. The socket bolts reside in shallow slots each side of chassis hole for 300B. In time, bolts
could sway out from slots to leave a dangling socket. But Jadis didn't see the stupidity of
not ensuring bolts were through a hole, and not in a slot right near hole edge.     


Power trannies on front of chassis are huge. They did get a bit warm though, maybe Bac is a bit high.....
But now these amps can sing well. Notice that Jadis OPT badges are missing from tops of new OPT
pots. Turned out Jadis pots had slightly concave tops, and badges were also dished, but new pots
have slight convex shape so brass plate badges would not lay flat nearly in silicone bed like on
original pots. In any case, screwing old badges down using 4 small c/s M2 screws meant  telling a lie.
Content inside pots wasn't made by Jadis. 


Two Hammond SEA1627 sit on my bench showing bell-ends have been removed from one OPT,
and replaced with aluminium fixing angles prior to potting with mix of dry sand and casting resin.
Sand was baked to remove moisture, then a 50-50 volume mix was used. Sand settles in
liquid resin, so more is added to bring submerged sand level up to minimize expensive
resin. The resulting concrete does not shrink much after it has set, so steel
sides of pot do not come loose and buzz at AF from stray magnetic field. But where any
buzz does occur, it is possible to drill a deep 5mm dia hole beside inside of 4 pot sides,
and then pour a second small volume of resin only which should run out into gap
between concrete and pot side.


Close up of fixing angles on OPT before potting.


Another close up of OPT before potting. The assembly dropped into pot with sliding fit
and the M6 screws pointing upwards. Large dia washes with 8mm holes were bolted down on
M6 screws, so that the larger angles were pulled up flush with the edges of the pot.
OPT was then slid to centre position, and potting mix poured, poked, pushed, and
cajoled into filling through just enough pouring gap between OPT and pot.
After a day, 2 x M4 c/s screws were put through pot to vertical angle.
The pot cannot come free.


A pic of faulty Jadis OPT and a Hammond SEA. After weighing the newly potted Hammond
I found it was about 1Kg heavier than the Jadis potted OPT, and I estimated the Hammond
has a higher core weight. Pot size I happened to have laying around was 10mm higher than
Jadis pots, but 5mm less for each side size, so internal volumes were near equal.
Don't ask me what exactly is actually inside the Jadis pot, but my careful and repeated
measurements indicate there is maybe a PP OPT which has fully interleaved laminations,
or C-cores with no air gap. Hence it would be lighter weight. But if there is a CT, it has not
been brought out to amp. If there had been a CT, I think the SE amp could have been
easily converted to a PP amp, using an extra 6SN7 twin triode for which there is plenty
of room where tubes are located. Class A PP 300B probably sound better. I make no
apologies for saying that, but really, its almost impossible to make a bad sounding PP
Plus, PP operation gives the slight benefit of class AB, and 25Watts is easily done,
and OP tubes can be biased at only 20W Pda each.

There are TWO spare transformers from a real pair of what were Jadis amps. They really do exist.
I ain't makin' up this story. Probably, the Jadis quality control guy fucked up big time on these 2 amps.
It may only be these two amps, and no others, but I suggest anyone who has Jadis SE300B amps
should have them tested by someone competent to find out if OPTs are crook. Of course, finding a
competent bloke on this sort of thing IS DIFFICULT, and he won't be your audiophile friend who
"seems to know a lot".

The owner is now very happy with my work. He was using Atma-Sphere OTL amps with their
supposed gloriously transparent sound, and "fabulous reliability", ( Never mind so many are
parked due to repeated bias failures.) Well, after a few nights with my reformed amps,
he has sold his Atmas-Sphere amps. Hey hey, another myth just hit the dirt !!!!

Do you realize what some audiophiles say to each other? Well, one guy tells another
some of the OP tubes in an Atma-Sphere can be removed if there is too much volume
because the amp is too sensitive, or in fact the source level is way too high so gain control
works near the bottom of its range. Less OP tubes means less volume, no? makes sense, yes?
Well no, because all tubes in all OTL are working under duress and in class B and they
overheat with alarming regularity. Using less tubes mean even lower levels of sound
cause remaining tubes to cook to death. Less tubes don't give less volume because levels
in OTL amps are subject to large amounts of loop NFB, so fewer horses are flogged
harder to maintain speed.

Well guy listening thinks, ah, I can remove one 300B to make it quieter. Hmm, turns out guy
has a Wadia CD player with incomprehensible manual instructions for its digital attenuator,
which probably sounds no better than a good 20k log pot, or a DACT switched attenuator,
all despite the Wadia sales BS.

I had to insist that the guy never ever remove any OP tubes in any amp at any time ever
lest he enjoy another huge repair bill and bad sound.

Folks, don't do anything I wouldn't do.

And don't do anything I would not enjoy.

Happy soldering to all.

To Re-engineered amps

To Educational and DIY directory

To Index