• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Mullard 5-20 KT88 PP blocks!

I thought I put the gate stoppers in the schematic. Actually I did, just never made a new pdf. Here you go.

George, what kind of current in the FET's? Even at 3mA, I calculated it would need a heat sink in this configuration.

Good word "Effortless". It explains my experience when playing with some breadboard SET amps a short while back with and without a low impedance drive.

hi SGregory,

for biasing the gate of the mosfet such as an IRF820, you will need a small dc supply, about 10 volts, which you can derive from the 6.3volt ac filament wiinding.....

you can use a 1meg gate bias resistor and small coupling caps from plate of the LTP, 0.01 is good....

i am sure SY agrees with this one...:D
 
Tony,
I am not sure what you mean. The bias would be set by the Vgs and will be somewhere around 4V from the Vs. The overall FET bias would be taken from B- rail since it would definatley be a negative value. In AB2, the FET bias point would set the kt88 bias via Vs since it is after the coupling cap. 1Meg bias may be possible but I have not calculated what HF roll off would be with the Gate capacitance at that value. (*It would make the LTP load even better) Per Eli we want it at least 470k based on his success of the El-Cheapo.

Help me if I am wrong with the above.

AB1 is a little different in that it is directly coupled to the LTP. In that case the bias is set by the plate of the LTP. The FET source is then cap coupled to the kt88.

Difference between AB1 and AB2:

In reality, I don't care if AB1 is used or AB2. I personally like the AB2 operation since it provides the possibility to drive the kt88 into positive grid with out having to deal with the blocking distortion. I am not sure if the LTP at these voltages has enough swing to get to into AB2. Eli please chime in here. I am not fluent enough with the math to manually calculate the swing and my simulator only has a resistive tail.

My reasoning for liking the ability to drive into positive grid is based on a simple bread board SET amp I built a short while back. It is not scientific. The purpose was to prove to myself the need for more drive. In this case I used a 6C45 as a driver and a 300b output into 5k. I initially had it cap coupled and later put George's powerdrive into it. (I didn't want sand in the signal path initially). The sound was night and day particularly with dynamic music.
I should set the amp back up with a kt88 and try both methods.
At 60W output, AB2 is less of an issue unless your speakers are of lower efficiency as there is a BIG difference between 6 Watts and 60 Watts. I suspect however, that even at 60W, there are transient spikes that could benefit from AB2 operation.

IMHO this will make the amp stand even further above the standard Mullard 5-20 from what it is today.


We should decide and get one of the designs off of the table so the build can continue. Tubemak, Eli, Boywonder, TAJ, and others, What say you?
 
Last edited:
The bias would be set by the Vgs and will be somewhere around 4V from the Vs.

yes, you are right about Vgs....between 2~4volts is about right, therefore your biasing arrangement must connect to a +voltage source instead of ground, then you can use a gate leak of about 470k to 1meg from the center of the pot to gate....this gate leak must be at ac ground, nessecitating use of a filter cap.

i am commenting on your latest AB2 drawing....
 
Last edited:
Actualy the gate is tied between B- and ground on that schematic. Therefore the gate bias voltage will be between B- and ground depending on the voltage divider. I am estimateing it should be able to dial Vg between a range of -70V to -45V depending on how we set up the divider.

We would have to do this for both AB1 and AB2. AB1 would as low of a total resistance as possible whereas AB2 we want the opposite by having as high of a total resistance as possible.
 
Both the source and the gate( through voltage divider) are tied to B-. The drain is to B+. The FET will pass more current or less current to satisfy the required intrinsic Vgs. Vs being established by the source resistor and the current that is passed across it. The value for the source resistor is determined around the desired current through the FET at the desired operating point.

IRF820 won't handle the voltage. The FET in consideration here is IRFBC20. If higher voltage >600V is needed can use a 2SK2700


It works the same way for both AB1 and AB2
 
Last edited:
Some remarks in no particular order.

As George (Tubelab) said, DC coupling the FET to the LTP plate does nothing about blocking distortion, "All" it does is: keep LTP gain up, keep LTP distortion low, and allow the use of a very comfortably sized KT88 grid to ground resistance. Those highly important gains are achieved, without introducing any complexity to the amp's PSU. :D Let's take George's 10+ mA. ID comment to heart. A 27.4 KOhm Mills MRA12 (pay the man his money), as the source follower load, does the job.

If you beat the snot out of a pair, 100 W. can be squeezed out of KT88s easily enough. IIRC, a B1 condition set will achieve that result. I would want 150 W. rated O/P "iron" if AB2/UL is the condition set. For the 100 W. trafos we've been considering, it makes sense (IMO) to use AB1/UL or AB2/triode.

Amps and speakers have to be viewed as parts of a whole. The high power this project will produce, including triode mode, is NOT to mated with high efficiency horn speakers whose driver cones are very light weight. Acoustic suspension, bass reflex, and transmission line stuff is the likely mate. A refurbed pair of AR3s, AR9s, Paradigm Studio series towers, big ProAcs, etc. are candidates. Speakers of that sort need to be mated with good amp damping factor. While no GNFB triodes are fine for horns, they lack the damping factor needed in combination with "typical" medium-low efficiency speakers. A small amount of loop NFB (say 5% of the O/P voltage) is, IMO/IME, needed, when triode mode "finals" are employed. NFB is a tool, not a crutch. It works well, when the open loop characteristics of the circuitry are good. Bludgeoning crappy circuitry with gobs of NFB is something I abhor. Guano can not be processed into Gold!
 
Eli, another problem with the AB2 as drawn is I don't think a 27.4k source resistor will work at 10mA unless we drop B- way down. I calculate that it has to be around 3k in order to set the kt88 bias even with a -90 rail. At 3k I suspect that it would be too low to keep the FET happy. Could use a current source under it but now it is getting too complex. The other option is to reduce the current but I don't see that is viable for sonics based on George's comments.

Or is my math wrong?

If my math is right then the AB1 is the way to go for both this reason and the OPT issue you bought up. MOST Importantly for simplicity sake.

Your comment on NFB is spot on. Horns need not apply unless your listening room is the size of a stadium.
 
Eli, another problem with the AB2 as drawn is I don't think a 27.4k source resistor will work at 10mA unless we drop B- way down. I calculate that it has to be around 3k in order to set the kt88 bias even with a -90 rail. At 3k I suspect that it would be too low to keep the FET happy. Could use a current source under it but now it is getting too complex. The other option is to reduce the current but I don't see that is viable for sonics based on George's comments.

Or is my math wrong?

If my math is right then the AB1 is the way to go for both this reason and the OPT issue you bought up. MOST Importantly for simplicity sake.

Your comment on NFB is spot on. Horns need not apply unless your listening room is the size of a stadium.


The 27.4 KOhm load resistor is for the AB1 setup, with gate DC coupled to 'T7 plate.

I've made no attempt to run any numbers for the AB2 setup with the source DC coupled to KT88 g1. For argument's sake, let's credit your 3 KOhm value with validity. That's no problem, whatsoever. A very crude guess for the O/P impedance of the source follower, when ID = 10 mA., is 100 Ω. Do you really think there will be trouble?
 
I've made no attempt to run any numbers for the AB2 setup with the source DC coupled to KT88 g1

If you have a -90 volt supply, and we assume a bias of about -50 volts (from the UL curves for for a KT88 with 450 volts and 50 ma) we need to drop about 40 volts across this resistor. For 10 mA we need about 4K ohms. I run my EH KT88's a bit hotter and I use two 10K resistors in parallel. The higher current does lower the output impedance a bit. I also tend to build amps with higher than normal plate load impedance to lower the output impedance, trading off a bit of output power.

I would want 150 W. rated O/P "iron" if AB2/UL is the condition set. For the 100 W. trafos we've been considering, it makes sense (IMO) to use AB1/UL or AB2/triode.

I have a pair of 400 watt Plitron toroids so the OPT shouldn't be a limiting factor for anything that I can plug into a conventional wall outlet!

Yep. I can't claim to have originated it. But I am listening to it as I type this :)

PowerDrive does wonders for transient behavior even if you never intend to beat the snot out of your output tubes. My Tubelab SE is still running the same pair of NX-483 output tubes that I pulled out of an old radio about 5 years ago. They are now 80 years old! Do you think your computer, cell phone, ao anything electronic will be useful in 80 years.

Amps and speakers have to be viewed as parts of a whole. The high power this project will produce, including triode mode, is NOT to mated with high efficiency horn speakers whose driver cones are very light weight.

I agree. The speakers that I am using are Yamaha NS-10M Studio monitors. They were very common in recording studios for their ability to reveal flaws in the mix. They are good for finding flaws in the amp too. They are NOT known for being tube friendly. Some amps like the Simple P-P (pentode mode) need a bit of GNFB to tame these speakers. I use UL with cathode feedback on my Simple SE with these speakers. The same amp works best in triode with zero feedback on Lowther based horns.

The KT88 P-P amp that I have been experimenting with does make big power in AB2 triode. As I mentioned in the long thread where it is being designed, it makes the most power with a 2500 ohm load, but sounded better with a higher load impedance. The trade offs between power output, load impedance and feedback level depend heavilly on the speakers that will be used. There are generalizations that can (and have) be used, but sometimes you need to tweak things with the intended speakers attached.
 
100R o/p impedance certainly meets the drive requrements of the kt88. Your point is valid, looking at the curves for the FET I can't see were the source resistor at 3k would be out of bounds. The only issue would be that Vds is close to its max in the AB2 config.

In AB1 the Vds would not be an issue.
 

taj

diyAudio Member
Joined 2005
The speakers that I am using are Yamaha NS-10M Studio monitors. They were very common in recording studios for their ability to reveal flaws in the mix.

Sorry TubeLab, with all due respect, and I'm not trying to stir things up, but that's NOT why we used NS10M's. We used them because they provided the same limitation in bandwidth that typical street 'boom boxes' had in those days (my days), but without any specific colouration of their own that would affect a mix too much, as compared to the same music emanating from the full bandwidth studio monitors. They are fairly flat, but not good sounding speakers and were only used as a reference for crappy speakers (again, without too much added nonsense of their own.) And we typically had to modify them to achieve even that (a tissue over the tweeter was almost universal). We used them to make sure a mix still had the instruments in the correct proportion when Joe Six-Pack played the music on their garbage blasters. I suppose that could be a rough definition of "revealing flaws in the mix", but only in relation to pop music being played on cheap mini-systems of the day.

Any music that needed to sound good on decent speakers would be monitored on the "big boys". You would not use NS10M's for mastering. They are not used much by serious studios anymore due to the change in public listening habits. (Ear buds and iPods would serve that purpose these days). Don't tell that to the snake oilers on eBay though, or their customers.

Because the size of most studios is vastly smaller these days (i.e. DIY musician-owned home studios), most mixing is done on high-end near field monitors like Adams, Genelecs, etc.

..Todd
 
Last edited:
There are a whole lot of conflicting stories about the reasons that the NS-10 series became so popular, but many people didn't like to actually listen to them. Some are here:

The Yamaha NS10 Story

http://www.soundonsound.com/pdfs/ns10m.pdf

I agree with your statement that recording studios try to emulate the popular playback hardware of the day. I was surprised to find a piece of plywood with two 6X9 car stereo speakers behind the mixing console at Criteria Recording (Miami) back in about 1970. Of course there were other speaker systems that could be switch selected. The engineer explained that much of the listening audience would be playing his work on an 8 track deck in their cars, so he wanted to hear the mix through a similar setup.

There are at least 3 different versions of the NS-10 series. The original NS-10's were meant to be used vertically. The NS-10M's were meant to lay on their sides. The NS-10M Studio had a different tweeter and crossover network to eliminate the need for the tissue paper!

Because the size of most studios is vastly smaller these days (i.e. DIY musician-owned home studios), most mixing is done on high-end near field monitors like Adams, Genelecs, etc.

I built my own home studio without the high-end budget in about 1977. I did have a nice Teac 3440 tape deck, but my speakers were home made using 6X9 car speakers, so was my solid state amp and all of the electronics. The room is small, 10 X 11 feet and it is also my lab, so it is quite crowded. Upgrades have come over the years, the Teac was sold and I now use Cakewalk Sonar and an EMU multi channel interface. Somewhere in the 1980's, I decided that the old home made speakers didn't cut it any more. I needed something that would fit the room and the budget ($500 max). Armed with several CD's of popular music (mostly rock) and a few that I recorded myself, I went speaker shopping. I hit the typical HiFi shops and the music stores. Sam Ash music had a mini studio with a wall of speakers to try, including Genelec, Tannoy, Yamaha, Peavey and others. My daughter played multiple instruments at the time which were purchased at Sam Ash, so they were cooperative on pricing. After several trips to their studio often accompanied with other musicians, I decided that the NS-10 Studio's sounded more like what I heard live than the other speakers in my price range. This was most obvious on the recordings of my daughters percussion ensemble. The fact that I got them for $300 sealed the deal.

Granted there is zero sound output below 70Hz even though you see the cone moving, but I have been able to tweak the feedback compensation cap on these speakers by sound alone (female vocals and cymbals) and then verify the results with square wave testing. Yes, I have heard several much better sounding speaker systems, but none of them fit my space and budget requirements.

I mentioned previously that I had constructed a KT88 P-P amp that used an LTP driver (two of them actually). I have been requested to post a link to that thread. It was started by someone else, and wasn't intended for KT88's even though that is what is currently in my amp. The (very long) thread is here:

http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp.html?highlight=6L6GC+AB2
 
I guess i'll chime in.

I'm not sure the current proposed amp is what I want to build. It all sounds very interesting, but I feel it has transformed into another amp altogether. I wish I had waited further into my studies to start the project so that I could better understand the issues involved. The amp seemed to had been almost done before the issue with the 12AT7 operating points. Is it not to late to go back to Taj's last schematic and talk about moving another triode into the LPT slot such as one of the "super Triodes"? Wasn't Johan looking into this? would this not be simplier?

That said, everyone should build what ever makes them happy. I now have enough to beat something out over time, and with more study.

I'll let boywonder, Taj, and the others that wanted to build input their thoughts.
 
Last edited:
Said once, said again, having the 'T7 as part of the small signal complement makes for a SUPERIOR net HD spectrum. We've worked the 'T7 operating conditions, to near perfection, assuming 460 V. of B+ are available. Per SY's gain structure remark in the parallel LTP thread, an EF86 voltage amplifier whose anode load resistor is small, gets the ship safely into port. :) Instead of just a RC decoupling filter between the 460 V. point and the EF86 load resistor, change the RC network to drop only a few V. and put a LR8 regulator setup in between the decoupling network and the load R. You get to dial the exact voltage needed in and any PSRR doubts disappear. ;)
 
I guess i'll chime in.

I'm not sure the current proposed amp is what I want to build. It all sounds very interesting, but I feel it has transformed into another amp altogether. I wish I had waited further into my studies to start the project so that I could better understand the issues involved. The amp seemed to had been almost done before the issue with the 12AT7 operating points. Is it not to late to go back to Taj's last schematic and talk about moving another triode into the LPT slot such as one of the "super Triodes"? Wasn't Johan looking into this? would this not be simplier?

That said, everyone should build what ever makes them happy. I now have enough to beat something out over time, and with more study.

I'll let boywonder, Taj, and the others that wanted to build input their thoughts.

Okay, I agree 120%. I love the 5-20 idea and we are getting way off mark. Let's get some final stuff decided and get going. I received my Hammond 100w transformers today, they are monsters! I have PT that I'm pretty sure will work ok. What's the rest of the story??
 
Said once, said again, having the 'T7 as part of the small signal complement makes for a SUPERIOR net HD spectrum. We've worked the 'T7 operating conditions, to near perfection, assuming 460 V. of B+ are available. Per SY's gain structure remark in the parallel LTP thread, an EF86 voltage amplifier whose anode load resistor is small, gets the ship safely into port. :) Instead of just a RC decoupling filter between the 460 V. point and the EF86 load resistor, change the RC network to drop only a few V. and put a LR8 regulator setup in between the decoupling network and the load R. You get to dial the exact voltage needed in and any PSRR doubts disappear. ;)

Thanks Eli, I'll look over those changes.