diyAudio

diyAudio (http://www.diyaudio.com/forums/index.php)
-   Tubes / Valves (http://www.diyaudio.com/forums/tubes-valves/)
-   -   KT88 Parallel Push Pull (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull.html)

6L6 28th May 2012 07:31 PM

Quote:

Originally Posted by richwalters (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull-post3040061.html#post3040061)

Here is my 250+250W Williamson...

Holy cow! Wow! That's amazing!!!!!! :D :D :D :D :D

Alexontherocks 28th May 2012 07:34 PM

@richwalters Very nicely done! Great PtP wiring!

es345 28th May 2012 08:02 PM

@richwalters

Are these KT90 from EI?
I had some reliablity issues with KT90 from EH and changed to KT120.

richwalters 29th May 2012 04:51 AM

2 Attachment(s)
Quote:

Originally Posted by costis_n (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull-post3040082.html#post3040082)
Care to elaborate? I have never met this term...

In short, I regulary use this earth matrix concept for power amps with AC transformers & chokes in close proximity, basically to avoid circulating currents in the chassis that can give rise to miniscule AC potentials between differing earth points, and since the noise figures of audio amps requires very low hum figures, a safe way is to use a busbar earth internally which avoids chassis potentials. The "earth matrix" loop is lifted off via insulated standoffs and earthed at the furthest point away from the power components at the input stages, (shown as crosses) roughly equicentral if stereo stages are being constructed. Obviously if a mono block is used this is considerably smaller.

See pic, follow yellow line which behind this is the copper bus. This one is solidly constructed with 4mm copper wire and this requires a hefty soldering iron. Many amps in the past used 2.5-3mm copper wire. The photo is another amp built on exactly the same lines as the previous 250+250W.

Strictly really one could do away with a ally chassis as used here, but since I use a 1kW switched mode power supply, this requires every particle of shielding and supression available, and the earth matrix is a pretty reliable method.

Metalworking tips:- Also shown is the brass lid for the transformer/switchmode unit; 0.5mm brass sheet from a hardware store, plus a hardmetal scriber, use a flat surface sheet and clamp metal ruler where desired bend is required and repeatedly keep scribing in the same grove on the inside until it can easily bend without breaking. Since it is thinner at this line, the inside can be reinforced by soldering.

richy

richwalters 29th May 2012 05:14 AM

1 Attachment(s)
Quote:

Originally Posted by es345 (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull-post3040148.html#post3040148)
@richwalters

Are these KT90 from EI?
I had some reliablity issues with KT90 from EH and changed to KT120.

EH KT90's used. I have found them consistent in my studio system. A friend was using KT120's in his and he wasn't happy..that makes the choice rather awkward as no others are suitable. I wouldn't dare run todays KT88's in UL at 600V.
The two power Williamson concept 500W amps shown left in pic have lower power saving settings 20W;50W for easy listening at 200V & 300V B+; For my bass guitar and octave shifter work, 600V at 65mA Iquies per tube is the norm without problems.
This system is capable of not only destroying ears and vibrating any socket loose but also striplights in arenas. This is the reason that the internal construction of these amps does require careful point to point work with generous soldering and support. Hence the input tube clamps. The loudspeakers too require internal bracing and this must be done correctly.

Keep listening up !

richy

costis_n 29th May 2012 08:14 AM

Quote:

Originally Posted by richwalters (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull-post3040716.html#post3040716)
In short, I regulary use this earth matrix concept for power amps with AC transformers & chokes in close proximity, basically to avoid circulating currents in the chassis that can give rise to miniscule AC potentials between differing earth points, and since the noise figures of audio amps requires very low hum figures, a safe way is to use a busbar earth internally which avoids chassis potentials. The "earth matrix" loop is lifted off via insulated standoffs and earthed at the furthest point away from the power components at the input stages, (shown as crosses) roughly equicentral if stereo stages are being constructed. Obviously if a mono block is used this is considerably smaller.

See pic, follow yellow line which behind this is the copper bus. This one is solidly constructed with 4mm copper wire and this requires a hefty soldering iron. Many amps in the past used 2.5-3mm copper wire. The photo is another amp built on exactly the same lines as the previous 250+250W.

Strictly really one could do away with a ally chassis as used here, but since I use a 1kW switched mode power supply, this requires every particle of shielding and supression available, and the earth matrix is a pretty reliable method.

Metalworking tips:- Also shown is the brass lid for the transformer/switchmode unit; 0.5mm brass sheet from a hardware store, plus a hardmetal scriber, use a flat surface sheet and clamp metal ruler where desired bend is required and repeatedly keep scribing in the same grove on the inside until it can easily bend without breaking. Since it is thinner at this line, the inside can be reinforced by soldering.

richy

Looks like a ground plane, without the plane. I ll keep it in mind, thanks.

es345 29th May 2012 05:41 PM

Quote:

EH KT90's used. I have found them consistent in my studio system. A friend was using KT120's in his and he wasn't happy..that makes the choice rather awkward as no others are suitable.
Interesting, my experience with KT120 is more positive. And in AB2 Mode I have got 200W out of a pair (UL mode with separate G2 winding)

richwalters 30th May 2012 03:57 AM

Quote:

Originally Posted by es345 (http://www.diyaudio.com/forums/tubes-valves/202546-kt88-parallel-push-pull-post3041340.html#post3041340)
Interesting, my experience with KT120 is more positive. And in AB2 Mode I have got 200W out of a pair (UL mode with separate G2 winding)

The problem is more field time results are needed vs. operating conditions.

With NewSensor I was badly letdown by their TungSol 6550's New Ed (coke bottle) with 2000 hrs being the end date. 470V 70mA quiesAB1; I believe others also spotted the fact that this class of tubes simply don't like being pushed over the 33W dissip ballpark, contrary to the data sheets. Many users also mention that the 6550 is part-power classed to the KT88; how wrong it turns out to be.
However, I remember in 1967 when KT88's were used in UL amps at 600V with a quies at 60mA without any fuss. Worst still, many designed amps were put in racks and it was the question of electrolytic cap reliability which was the issue, not tubes. That has now turned around as electroytics in optimum positions are lasting far longer. I wouldn't operate in those conditions now.

I turned to the EH KT90, which in some ways behaved how an original KT88 used to be, i.e take punishment. I had thought about the EL156 group, but again not enough field tests with the chinese octal version and this tube is pricy with no second source.

richy

Xpersephone2 14th September 2012 01:33 PM

Progress progress
 
6 Attachment(s)
Progress progress, when Iím not working or being married.
The driver stages have now been put onto a board and tested; I've posted the hopefully final optimised version and a picture of the board. The screen shot of my scope shows the beginning of obvious clipping (Ďbottoming outí) from the board at about 155V p/p, given 500V B+ on the KT88íd bias will be about -57V so thatís plenty. More to the point at 100V p/p THD of the driver is less than 1% which will be swamped by the output stage.
Iím going to regulate the B+, and hence the bias as well. Iíve tested the B+ regulator posted based on the Gary Pimm / Swenson circuit at 500V 500mA out; there are some changes from the Gary Pimm design because I could never get it stable. Iím going to use a Hammond 379WX for the B+, bias and KT88 heaters and a Hammond 266N12 for the driver heaters. Iíve modelled the dissipation in the MOSFETs at around 60W using the 425-0-425V taps on the Hammond so Iíll be using a 0.2C/W Fischer forced air heatsink but Iíll run the fan at a lower than rated voltage to keep it quiet. Iím going to earth the heatsink with the MOSFETS on insulators and a fuse between them and the 850W transformer in case the insulation pad ever fails.
Iíve modelled a negative version of the Gary Pimm / Swenson as shown on the screen shot of LTSpice and Iím waiting the MOSFETs to test it. The driver stages will have DC heaters and Iíve tested the attached design using Linear LDOs from a 6.3V heater.
I will also need a 12V PSU to drive the fan for the heatsink and Iím going to use that to run some protection circuitry (soft start on the KT88 heaters and bias protection) so there will be five PSUs to worry about.
I imagine it will take me six months to get all that on another board and test itÖ

richwalters 14th September 2012 04:17 PM

2 Attachment(s)
Xperse

Different avenues;
I've gone Williamson diff with 12BY7A's; which has active pull down to deal with Miller effect; o/p voltage swings same with B+ only 360V, tube currents higher; o/p Z around 2K.....all kinds of comparisons can be made. 20dB gnfb

See driver pic: This circuit can handle a 1nF cap connected to each output per Miller simulation; somewhere I'll dig the squarewave graph out.
At the moment I'm in the process of converting and copying all old drawings from 32 bit to my new 64bit machine: Not everything goes quite as it should do.

Looking at your grid leaks; the effective output AC resistive shunt load on each cath follower driver is a little higher than the driver cathode resistor value of 33K.
With the Williamson I've kept this ratio wider; so with this active driver stage, it shunts the Miller-effect better. Becareful with this type of stage as the bandwidth is considerably higher than the cathode follower; thus the Miller effect sees the 33K resistor in series; and can only be "pulled down" the value of this cathode resistor. My Williamson has -3dB point at 70Khz ...but as you know it all depends on the o/p tranny quality.

I've posted the spectrum at 150W o/p.. it's clean. The other which I'm working with is at 250W o/p, at slightly higher thd, onset clipping.
KT90's at 600V, will parallel p-p will reach this power region; there is enough drive.
Expect 20% less power with 88's. Some makes of these won't be happy at 600V...so I'm with KT90's.

The remainder of the story is with the power supply. For this o/p power, my dual/stereo amp needs stabilised 600V at 1kW rating reserve. This contains an active PFC boost converter and is beyond forum work.

HINT; If one doesn't to see switchmode power supply HF harmonics in the audioband spectrum, i.e lower than -80dB down; a complete understanding of EMI issues and magnetics is vital.
Dunn'a want to do this way ? Stay with simple passive components.

If one tabulates and documents the results as one designs this; you have the silk purse in understanding it all. Alot of work, but EXcel and other programs are worth it.

richy


All times are GMT. The time now is 11:56 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio

Wiki