[Phil Allison[_3_]]

** Hi all,

a neatly built, home brew SET amp arrived on my bench this week, I
understand it is about 10 years old.

The tube line up is two 300Bs in parallel and one 6SN7 per channel, the HT
is 550 volts and each 300B has its own grid bias adjustment. There is no NFB
used and the PSU has diodes and a metal bridge for DC heaters on the 300Bs.

One of the output transformers ( UBT-1s by One Electron ) was open primary
and two of the Chinese 300Bs were stuffed.

I have one good channel and I so did few tests, grid bias set at -105V and
plate current is close to 55mA per tube.

Max power is 21 watts at 2.2% THD - mostly 3H.

At 10 watts, the THD is 0.85% - mostly 3H.

At 1 watt, the THD is 0.14%, mostly 2H.

The damping factor is 4.5

Signal to noise is -80 dB rel full power.

Input sensitivity is 260mV.

All the above is with an 8 ohm load on the 4 ohm tapping of the OT.

These numbers are a rather better than I expected.

Any comments ?


..... Phil

[[email protected]]

On Saturday, November 3, 2012 10:00:02 PM UTC-4, Phil Allison wrote:
** Hi all,



a neatly built, home brew SET amp arrived on my bench this week, I

understand it is about 10 years old.



The tube line up is two 300Bs in parallel and one 6SN7 per channel, the HT

is 550 volts and each 300B has its own grid bias adjustment. There is no NFB

used and the PSU has diodes and a metal bridge for DC heaters on the 300Bs.



One of the output transformers ( UBT-1s by One Electron ) was open primary

and two of the Chinese 300Bs were stuffed.



I have one good channel and I so did few tests, grid bias set at -105V and

plate current is close to 55mA per tube.



Max power is 21 watts at 2.2% THD - mostly 3H.



At 10 watts, the THD is 0.85% - mostly 3H.



At 1 watt, the THD is 0.14%, mostly 2H.



The damping factor is 4.5



Signal to noise is -80 dB rel full power.



Input sensitivity is 260mV.



All the above is with an 8 ohm load on the 4 ohm tapping of the OT.



These numbers are a rather better than I expected.



Any comments ?





.... Phil

UBT-1 is rated at 160mA on the primary... Did those 300Bs develope a dead short?

[Phil Allison[_3_]]

UBT-1 is rated at 160mA on the primary...


** Yep - the total idle current is 110mA.


Did those 300Bs develope a dead short?


** According to the owner/builder - there was a flash / bang / wallop event
and then a dead channel.

One dud 300B draws no bias current while the other draws waaaay to much, the
UBT-1 has an open deep inside but shows no sign of shorted turns.

What has me intrigued, however, is the low THD figures at and near full
power PLUS the fact that 3H dominates.

The test results are far more like a PP class A triode amp than a SE one.

The 6SN7 is running at full gain ( about 18 times per half ) and there is
definitely NO negative feed back - in fact the two speaker windings have
been left floating.



..... Phil

[flipper]

On Sun, 4 Nov 2012 16:13:05 +1100, "Phil Allison"
wrote:





UBT-1 is rated at 160mA on the primary...


** Yep - the total idle current is 110mA.


Did those 300Bs develope a dead short?


** According to the owner/builder - there was a flash / bang / wallop event
and then a dead channel.

One dud 300B draws no bias current while the other draws waaaay to much, the
UBT-1 has an open deep inside but shows no sign of shorted turns.

What has me intrigued, however, is the low THD figures at and near full
power PLUS the fact that 3H dominates.

The test results are far more like a PP class A triode amp than a SE one.

The 6SN7 is running at full gain ( about 18 times per half ) and there is
definitely NO negative feed back - in fact the two speaker windings have
been left floating.



.... Phil



Sounds like by either intent or serendipity the 300B and 6SN7 2'nd
harmonics are canceling.

[Phil Allison[_3_]]

"flipper"
"Phil Allison"

UBT-1 is rated at 160mA on the primary...


** Yep - the total idle current is 110mA.


Did those 300Bs develope a dead short?


** According to the owner/builder - there was a flash / bang / wallop
event
and then a dead channel.

One dud 300B draws no bias current while the other draws waaaay to much,
the
UBT-1 has an open deep inside but shows no sign of shorted turns.

What has me intrigued, however, is the low THD figures at and near full
power PLUS the fact that 3H dominates.

The test results are far more like a PP class A triode amp than a SE one.

The 6SN7 is running at full gain ( about 18 times per half ) and there is
definitely NO negative feed back - in fact the two speaker windings have
been left floating.



Sounds like by either intent or serendipity the 300B and 6SN7 2'nd
harmonics are canceling.


** Bingo !!

Seems an eccentric Russian gent was responsible for the circuit design.

The owner has agreed to purchase another UBT-1 from the USA and new 300Bs
and send them to me.

Then I can hook the beast up to my own speakers ( 10 inch 3 ways plus active
sub ) for a listen.



...... Phil

[John L Stewart]

Originally Posted by Phil Allison[_3_]

"flipper"
"Phil Allison"

UBT-1 is rated at 160mA on the primary...


** Yep - the total idle current is 110mA.


Did those 300Bs develope a dead short?


** According to the owner/builder - there was a flash / bang / wallop
event
and then a dead channel.

One dud 300B draws no bias current while the other draws waaaay to much,
the
UBT-1 has an open deep inside but shows no sign of shorted turns.

What has me intrigued, however, is the low THD figures at and near full
power PLUS the fact that 3H dominates.

The test results are far more like a PP class A triode amp than a SE one.

The 6SN7 is running at full gain ( about 18 times per half ) and there is
definitely NO negative feed back - in fact the two speaker windings have
been left floating.



Sounds like by either intent or serendipity the 300B and 6SN7 2'nd
harmonics are canceling.


** Bingo !!

Seems an eccentric Russian gent was responsible for the circuit design.

The owner has agreed to purchase another UBT-1 from the USA and new 300Bs
and send them to me.

Then I can hook the beast up to my own speakers ( 10 inch 3 ways plus active
sub ) for a listen.



...... Phil

There was an article on 2H cancellation in SE amps published about 10 years ago, I think in Glass Audio. Used the curvature of the driver to cancel curvature in the OP Tride. One would think this possible at only one level on the output.

At the time I thought that might give rise to 4H, Etc thru harmonic multiplication. Another article by an Australian I recall did something similar by paralleling the ouputs of two seperaste SET amps in antiphase (for the 2H). I recall sending a copy to Patrick T at the time.

Probably somewhere in my pile, if so I will pass it on.

The WE data book shows 3H as well as 2H in all SET amplfiers.

Cheers to all, John

[patrick-turner]

On Sunday, 4 November 2012 13:00:02 UTC+11, Phil Allison wrote:
** Hi all, a neatly built, home brew SET amp arrived on my bench this week, I understand it is about 10 years old. The tube line up is two 300Bs in parallel and one 6SN7 per channel, the HT is 550 volts and each 300B has its own grid bias adjustment. There is no NFB used and the PSU has diodes and a metal bridge for DC heaters on the 300Bs. One of the output transformers ( UBT-1s by One Electron ) was open primary and two of the Chinese 300Bs were stuffed. I have one good channel and I so did few tests, grid bias set at -105V and plate current is close to 55mA per tube. Max power is 21 watts at 2.2% THD - mostly 3H. At 10 watts, the THD is 0.85% - mostly 3H. At 1 watt, the THD is 0.14%, mostly 2H. The damping factor is 4.5 Signal to noise is -80 dB rel full power. Input sensitivity is 260mV. All the above is with an 8 ohm load on the 4 ohm tapping of the OT. These numbers are a rather better than I expected. Any comments ? .... Phil


I am presently working on a pair of really awful Jadis 300B mono blocs made in 2004 and each with 2 x parallel 300B output tubes, with one 6SN7 as the input with driver. Smoke, noise and silence brought them to my bench.

From what you are saying, you have a stereo power amp with a total of 4 x 300B, two in each channel. And one channel is dead because OPT primary is open, and two 300B, presumably in that dead channel are stuffed. They get stuffed real easy from even a short time being over heated. But it seems you have fixed bias. There should be a bias pot for each 300B. But with fixed bias, having B+ = Ea = 550Vdc and Ia at 55mA, Pda = 27.5 Watts which seems OK except that 300B don't much like having Ea that high, and the slightest bias bother, and POOF, they arc over internally and self destruct. Grid wires overheat and deform in shape, and typical failure involves grids going to say +400Vdc, and Ia rising to saturation and then loss of emission, and maybe shorting and fusing of OPT primary, all because dumbclucks who made the amp don't understand the need for bias failure protection of fragile 300B. The anode load for 300B can be calculated as RLa = (Ea / Ia) - 2 x Ra and as Ra is maybe 1,200 ohmns at the bias point then RLa should be = 8k6, and PO = Ia squared x RLa / 2 = 10.75 Watts and OPT losses reduce this to about 10Watts, and with TWO 300B you get the 20Watts, providing each 300B hasn't been partially wrecked a bit as they so often are by the time the amp is serviced. Its not uncommon to find grid bias voltage needed to get the wanted Ia to be 40% different, and to find that ac signal current to the OPT is 30% different, which tells you tubes are not matched, and about as different as say 6L6GC and EL34 in triode. With 2 tubes, the OPT would be 4k3 : 4,8,19, and Lp with Idc should be at least 34 Henrys. You should measure the good OPT to make sure these figures are about right. Now if some Phuctard connects 4 ohms to the 8 ohm labelled output and he turns up the volume, it ain't too good for 300Bs. Its not uncommon to find that because signal ac currents are so different, that one tube "sees" say 10k and the other sees say 5k, and average is 4k, and PO is then usually less than if both tubes are matched within 10%.

Many asian made SE amps have extremely badly concieved load matching, as if the makers were merely copying the very worst of what was made in 1955 in US or UK etc. Many DIYers just don't know about load matching, loadline analysis, Ohm's Law, or anything else, but they continue making great piles of ****.

I guess you don't know what caused the failure. But owners twiddling bias pots badly while confused, drunk, stoned etc, cause a heck of a lot of the profits earned by companies making 300Bs. Nevertheless I see there is a major problem with such a high B+. Now what is the B+ when you have say two channels each drawing say 120mA? if there is a tube rectifier, or two of them, typically 5U4, you'd find B+ will be maybe 450Vdc, and then if you find the mains tranny was meant for 220Vac mains and you have typical Oz mains at 245Vac, then B+ will sure be way too high. Chances are B+ is meant to be about +400Vdc max with fixed bias, but it has ended up much higher for a number of reasons and of course probably there are not say 4 taps on HT winding so B+ can be optimised to suit an available OPT load.

In the Jadis amps which I have totally gutted and re-wired I could get B+ of only +430Vdc at 140mA with Si rectifiers. I have re-used the available 2 x 9.5Vac windings to make 2 x 5V x 1.2Adc supplied to each 300B, and then used Rk & Ck = 1k0 + 94uF to each 300B which means Ek settles at about +70Vdc, Ea = 355Vdc approx. Anode Load is 1k8 for the 2 tubes which means each 300B sees 3k6, and PO = 2 x 8Watts = 16 Watts, all of which is less risky and more reliable than when there was just ONE 5V heater supply for both tube and just one Rk = 300 ohms. The original Jadis had 5U4 but Si diodes were found strapped across 5U4 tube diodes giving the B+, in a stupid effort to get high B+ and more than the feeble PO of 10W available from 2 x 300B for the 2 amps which had cost $19,000 new, but only $6,000 second hand. The Rk of 300ohms isn't high enough and Ia is too high and thus Pda was over 30Watts per 300B. With only 1 x Rk, if one tube inevitably draws more Ia than the other, it raises Ek which then reduces Ia of the other so I found 50% difference in Ia even with serviceable 300B tubes, one of which was at Pda of 37W. Jadis or whoever put Si dioes in didn't increase Rk. Then with the variations in Ia and load sharing one 300B would blow its anode fuse of 0.16Amps, and then leave the other tube with higher B+ and Rk way too low, and so it overheated and fuct up, and the mains fuse would blow and or its 0.16A fuse blew after the tube became damaged.

Nearly all 300B can be safely idled at PDa = 28Watts max, and one MUST be able to bias each individually, with a bias pot or with separate RK of about 1k0. There are a few makers of 300B which do allow 40W Pda continuously in class A, Emission Labs and KR Audio may make such, but one pays through nose. Most of the large tube matching differences due to match drift will be lessened to less than 10% with such separate Rk biasing. If such biasing is installed and Ek bias only rises to say +50V in one tube and say +70V in the other, then the one with low Ia is stuffed. Its a sign they are sick, even though they look quite OK. 2 of 4 EH300B tubes in the two Jadis began to measure poorly in the amp which had given troubles, and then which I began testing. Amp was fine with 2 good 300B, but with 2 partially fuct 300B it would not bias up Ia properly, and then they each began to get a short circuit arcing internally as soon as signal was applied and Eg1 would rise to +300Vdc. Its important to have separate coupling caps and biasing resistors where you have separate Rk because if one tube gets this failure of bias you don't want the +ve grid on one raising Eg1 on the other tube. I found that when Eg1 rose to +300Vdc, the Ek then slowly rose from say +50V just after turn on to about +135Vdc which meant Ia had gone to about twice the idle value, but still not enough to blow the 0.16A anode fuse. Pda is limited by the rise in Ek and slight drop in B+ when Ia is doubled. Having Eg1 rise so high causes no damage itself, and what causes damage is the Ia rise in OPT primary. Hence it is IMPERATIVE to have active protection to TURN OFF THE FUKKING AMP if one or more OP tubes has bias Ia rise to more than 1.5 x idle value for longer than 4 seconds, so you need diodes from each Rk and resistance divider to work an 106D sensitive gate SCR which latches on a relay which turns of the main large PT mains winding of the amp. Relay and SRC should be powered from an auxilliary 5VA 12Vdc supply with small separate mains tranny. I'm the only **** in the Universe with the foresight to ALWAYS install such active protection to STOP **** HAPPENING. After 15 years of making amps, several have had bias failures from failing OP tubes especially where owners have bought NOS EL34s only to find 1 or 2 of say 8 treasured Mullards have failed within a month, and that they should have bought a few spares.

In both the Jadis POS amps I have here, the two OPTs have NO AIR GAP. I could not believe this until I carried out tests 3 times, two different ways. With no DC in primary, Lp = 88H, but with only 120mAdc, Lp was 1H, and serious IMD occuured at all F when average PO exceeded 1Watt. THD at 1W at 50Hz was 5%, with THD rising exponentially above that because of the core dc reduced the iron permeability and lowered Lp to 1H. Used with cap coupling from a signal source to anode winding, response was OK because Lp = 88H. Because it was far too difficult to get existing OPTs out of their pots filled with epoxy, I decided to use Hammond SEA1627 OPTs which have Z ratio 2k5 : 4, 8, 16. I measured Lp at 17H with 140mAdc, and with Vac at 10Vrms, 50Hz, a bit lower than the Hammond spec, but OK because ideal loading is 1k8 so that best loading at sec for max 16W is 2.9r, 5.3r, and 11r. This means that the "8 ohms" tap on sec is ideal for 5.3r, but any load between 3 and 11 ohms can in fact be used and one gets at least 10Watts of useful PO. I have applied 10dB of GNFB and altered the driver amp and specs I now get are far better than anything done by stoopid Jadis. Getting unconditional stability with Hammond OPTs wasn't easy, and I had to employ several phase tweaker R&C and an OP zobel across the "16r" sec winding. But this means that full PO bandwidth is 22Hz to 50kHz, THD 2% at onset of clipping at 16 Watts. At -12dB, 2.25Vrms to 5 ohms, BW is unpeaked and 8Hz to 50kHz, quite OK for any amp. Rout = 0.6 ohms, THD negligible. I've used the SEA1627 Hammond OPTs before to replace walnut sized OPTs supplied in kits put out by Andrew Key in 1990s. Such 300B amps sound best with speakers with sensitivity at **60Hz** at over 92dB at 1W/1M. Many old speakers are rated for high sensitivity of say 93dB+ at say 1kHz, but by 60Hz their output has declined 10dB. This fact about many old crummy speakers whose suspensions have become stiff as a board have poor bass response. With so many ppl failing to use good OPTs and biasing etc, etc, etc, its no wonder 300B amps are renowned only for their midrange, and performance when only one violin or acoustic guitar is played in a recording.

I'm left with a couple of Jadis OPTs which are not very good for anything, unless they be used in a parafeed amp. I thought of converting the Jadis to parafeed but the choke needed was too big to fit anywhere, and I don't have the time to wind 2 chokes and I have to keep costs low. The alterative was to have 2 300B in series PP with +880Vdc supply with cap coupling to OPT. But I'd have needed one of Hammond's quite suitable 10k : 10k ISTs, and then put in an extra tube socket for EL34 in triode to drive it. Nup, its been better to just rip out existing OPTs and put in Hammonds. I happened to have two OPT pot cases same size as Jadis so bell ends can be removed from Hammonds and they will be potted so appearance will be the same with transfered Jadis name plates on top of pots, but the schematic sheet glued to bottom cover will tell the next owner what's really under the bonnet. I'm preparing a web page about Jadis experience and to give others a guide about best use of 300Bs.

Patrick Turner.