6S19P Amplifier

the thermionic watercooler

6S19P Amplifier

Postby sorenj07 » Wed Jan 31, 2007 9:24 pm

Here we go, official thread. These tubes caught my eye. At barely more than $3 per, they're certainly a bit easier of an entrance into decently powerful triodes without going down the route of the 2A3/300B etc.

Image
(two of mine)

Nothing definite so far, you can take a look at this tube's specs in the "tube 101" section.

so far I only have the tubes, 10 of the 6S19P, and two maybe-dead 6N1P's that I might try as the input/splitter stage.

B+ will probably be around 320V, with 220V on the plates and 50mA of current per. I haven't figured out exactly how to end up with this voltage and ~240mA of current, but I do have a 193M (10H, 300mA) and a pair of 6AY3B damper diode tubes, good for 175mA per BUT it doesn't say anything about a choke input filter, I suspect they'll be just fine.

The OPT primary impedance will probably be 2.5K (or 5K, I'm still confused). Power probably won't need to be more than 15 or so watts.

The driver stage is, to quote myself:

a single 12AT7 per channel to drive the 6S19P's. grounded-cathode at 300V (filtered down from 320V) into a 1/2W 33K plate load, with an unbypassed 100-ohm cathode resistor. 3mA plate current, and a voltage swing of -85 to +86.1. If I drop the cathode resistor to 90 ohms, I get -86/+86.1V. Unfortunately, the output impedance is pretty high at 12.1k...

Before that might be a 6N1P (probably not, because they're voltage hogs) or 6DJ8/6N23P doing input and phase splitting duty.

How does this sound so far?
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby sorenj07 » Sun Feb 18, 2007 11:50 pm

well, i've hit a couple snags. mostly, it's that this tube requires a lot of voltage swing on the cathode, but at the same time, i'm trying to work with a main B+ of around 280V. it's hard to feed a driver stage 280V or less, assuming more filtering after B+, and expect it to swing that kind of voltage.

For example, I've picked a bias point, given a 280V B+, at 200V and 40mA on the plate, at around the -75V mark. I quartered an 8K a-a output impedance on an output transformer to get 2K, which gave me my slope.

Depending on the power transformer, I might try to run the tubes VERY hot at, say, 60mA and around -55V on the cathode. A big 300-0 300-0 500mA Antek toroid into a 10H 300mA choke might do the trick. In that case, I'd probably only need to swing 65V in order to have decent headroom, which is a bit easier. I'm having trouble figuring out TubeCAD, though. I can arbitrarily set some parameters to get a pretty decent "Vo max" statistic, but what happens if the "calculated part value" for Rk doesn't match the one I have?

The long-tail pair is also pretty difficult to get working in TubeCAD... well, I guess I have plenty of time.

The plate dissipation I calculated (pretty simple, 160V on plates * 60mA current) is 9.6 watts. 19.2 watts of dissipation, but at what efficiency? I've got my eye on a the Edcor XPP15-8-8K, 8K a-a, 15W, at only $19 per :)
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby Shannon Parks » Mon Feb 19, 2007 6:45 am

Soren, if you want to use a LTP and have below 400V to work with, then definitely try using the LM234Z/LM334Z. Use the 'B' tab in TubeCAD and ignore the Rk and the B- VDC. You will need to size the current set resistor to make sure there's at least a 7V to 15V drop across the LM234Z, but otherwise, Tube CAD can help you sort this voltage swing out. To size the current set resistor, just look at the plate voltage and check out the tube curves in the datasheet. Hope all this makes sense. You can have a LTP at those low voltages, but you'll need some tiny help from silicon. Oh, you'll also need to AC couple this stage.
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby sorenj07 » Mon Feb 19, 2007 10:33 am

It's not making extreme amounts of sense but then again it's still really early for me, and I haven't gotten out one of my trusty datasheets. I do understand the part about AC coupling, though, there's no RC filter tacked on to the output, right? Thanks for the tips about 'CAD though, it definitely frees me of some doubts about the legitimacy of what I was doing... I'll be back.

On an unrelated note, I'm considering designing for a pair of monoblocks with the 6S19P in PPP, mostly because I'd rather have more than like 8 watts per channel, and the tube is pretty cheap and plentiful. 60mA per tube x 4 = 240mA :) pretty toasty!
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby sorenj07 » Mon Feb 19, 2007 10:39 am

Also, what percent for "balance" would be acceptable here?

here's what I've got so far:

Image
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby sorenj07 » Mon Feb 19, 2007 11:32 am

separks wrote:Soren, if you want to use a LTP and have below 400V to work with, then definitely try using the LM234Z/LM334Z. Use the 'B' tab in TubeCAD and ignore the Rk and the B- VDC. You will need to size the current set resistor to make sure there's at least a 7V to 15V drop across the LM234Z, but otherwise, Tube CAD can help you sort this voltage swing out. To size the current set resistor, just look at the plate voltage and check out the tube curves in the datasheet. Hope all this makes sense. You can have a LTP at those low voltages, but you'll need some tiny help from silicon. Oh, you'll also need to AC couple this stage.


Umm, just to be clear, where in this dummy schematic should I stick the SS current source? should I replace the tail with it or something?

Image

note: i figure the 100-ohm resistors in series with the 6S19P plates are to ensure equal voltage distribution, right? i saw something similar used in a PPP 6550 amp on Bonavolta
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby Shannon Parks » Tue Feb 20, 2007 7:04 am

Hi Soren,

First - I think we can cut out the first stage. This ends up a little like the PP 6EM7 amp Ed & I built. Very similar.

Actually, I was wrong about TubeCAD. Use the 'C' tab for the Long Tail in TubeCAD instead. The other tab gives you much higher Vout due to the negative rail. So the rules for using the 'C' tab and a LM234Z is this:
- You have to swag the saved voltage across the replaced tail/cathode resistor. In this application, it will be around 100V. So add 100V to your B+ voltage 'Circuit Variable'.
- This 'C' tab assumes a DC coupled circuit due to the high voltage drop across the tail. Now you don't need this DC coupling from a previous stage and the grid can just be referenced to ground. Simple.


Image

Now note the V bias voltage. It is only -2.5V, so we really need an additional 5 to 10 volts to make sure the LM234Z is turned on well. Since it would be nice to leave the grid at 0V, we need a negative rail under the LM234Z. We could use a shottky bridge diode off of the filament supply to get this easily. The currents are low and ths PSRR high, so a simple one cap filter and four diodes would do this. Hey, we are talking about ditching a whole stage at this point. :)

OK, simpler solution would be choosing a tube that wants a higher V bias:

Image

Hey, this 12BH7 is looking good - actually, I'm getting awesome voltage swing. How much do we need now for those Russkie tubes? The V bias means no negative supply is necessary. The gain is less, but since we are not using feedback, voltage swing is the important thing here.

Stammering and ramblin' here... :)
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby sorenj07 » Tue Feb 20, 2007 7:30 am

yeah a 12BH7 was suggested on DIYAudio as well and it's starting to look like a better and better candidate. Depending on where I decide to bias them, they'll want up to 75 or so volts of swing. A lot, I know, but it seems manageable given the numbers you've shown me :) Thanks for the help so far, this is becoming very interesting. Do you really think it'd be OK to drop a driver stage, by the way? I love the idea from a simplicity point of view but would it sacrifice headroom or something?

One other thing - what does the "balance %" statistic mean?

p.s. where does the voltage of 380 come from? my B+ can only be 280 or so. or are you suggesting a separate little power transformer for the 12BH7's? Given a big chassis, one of the hearty 500mA AnTek toroids could power 8 6S19P's at 280VB+ 60mA each, and something like a 270BX could take care of a 12BH7 per channel. It'd be a big chassis though for stereo - I'd probably want a fair bit of inductance on the choke input filter, and the monstrous 193Q is looking like the way to go...

It's amazing how far this design has come, and I hope to have a viable option before too long. Now, to convince someone who has some cash to build it... Yellow_Light_Colorz_PDT_07
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby Shannon Parks » Tue Feb 20, 2007 12:59 pm

sorenj07 wrote: Do you really think it'd be OK to drop a driver stage, by the way? I love the idea from a simplicity point of view but would it sacrifice headroom or something?


I forget if you rebuilt your linestage or did another from the ground up, but with a linestage and no negative feedback loss, I think why not? My 6B4G SET is around 16dB of overall gain.

sorenj07 wrote:One other thing - what does the "balance %" statistic mean?


This is calculating how matched the out of phase outputs are. Notice how it suggests a different value for Ra2. I think the CCS makes this much tighter and I wouldn't worry about it - leave the plate resistors the same for now.

sorenj07 wrote:p.s. where does the voltage of 380 come from? my B+ can only be 280 or so. or are you suggesting a separate little power transformer for the 12BH7's?


I initially swagged the CCS could gain you 100V - ie, 100V is not being watsed in a longtail. In reality, you are saving 50.7V (ie the grid voltage in ours will be 0V). This needs simulation needs a little reworking, but it shows what I'm talking about. Adjust it until the extra B+ V and the grid voltage are the same.
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby sorenj07 » Tue Feb 20, 2007 1:41 pm

separks wrote:I forget if you rebuilt your linestage or did another from the ground up, but with a linestage and no negative feedback loss, I think why not? My 6B4G SET is around 16dB of overall gain.

I ditched the 6SN7 linestage and used the parts to put together a pretty basic grounded-cathode 12B4-A linestage but I'll probably be selling it pretty soon to fund some musical projects I'll be taking on this spring, and/or to finish the little gainclone I need a chassis for, or the 6L6 monoblocks I've been "working on" since August. Good news on that, though, all film caps in the PSU! (2 80uF motor runs for the 1st LC, then a 40uF motor for the 2nd LC, then two dry-film 22.5uF's for the first two legs of the Williamson. the only lytics will probably be the 6L6 cathode bypass caps, and they might have to be BG's)

separks wrote:I initially swagged the CCS could gain you 100V - ie, 100V is not being watsed in a longtail. In reality, you are saving 50.7V (ie the grid voltage in ours will be 0V). This needs simulation needs a little reworking, but it shows what I'm talking about. Adjust it until the extra B+ V and the grid voltage are the same.
[/quote]
I'll get on that but I'm still a bit confused as to where the 100V are coming from, seeing as my power transformer will probably be something like 300 or 325V into a choke input, and then there would be another RC or LC filter after the main LC for the OPT... How can I "save" voltage that was never there in the first place? Does the CCS actually up the whole voltage?

The 280V is _before_ waste in a longtail, and before the plate load resistors, right?
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby Shannon Parks » Wed Feb 21, 2007 12:40 pm

sorenj07 wrote:I'll get on that but I'm still a bit confused as to where the 100V are coming from, seeing as my power transformer will probably be something like 300 or 325V into a choke input, and then there would be another RC or LC filter after the main LC for the OPT... How can I "save" voltage that was never there in the first place? Does the CCS actually up the whole voltage?

The 280V is _before_ waste in a longtail, and before the plate load resistors, right?


Your 280V B+ should be enough with the LM234Z. We are changing the B+ in TubeCAD to get a very similar circuit. When using the LM234Z, the only waste you'll have is the 7V to 10V you'll be dropping across it, versus a much higher voltage with a resistor, which could be over 100V. So with 280V B+ in your amp, you basically have a usable 270V to 273V for amplification with the CCS. A 380V B+ in a unit with a resistor for the longtail would be functionally equivalent. Think of a fixed vs cathode bias amp with the same power supply, and why does the fixed bias amp supply more power? Because we aren't wasting usable gain voltage at the power tube for the biasing.

In this application, it looks like we are only saving 50 or 60V, so it needs to be retested from my last sim. The 'Vg DC' will end up being 0V in your application. And you must make 'V B+' minus 'Vg DC' equal 280V. Sorry I'm not explaining this well. ;)
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby sorenj07 » Sat Mar 24, 2007 1:37 pm

Update! I'm probably going to use a 50VA, 9V toroidal power transformer as my OPT. The dual 115V primaries will act as a centertapped primary, and using 9V secondaries in parallel should yield me an impedance ratio of ~1306:8 ohms, which is perfect for the very low plate resistance of the 6S19P. I have no funds though so this is on the back burner :)
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby Shannon Parks » Sun Mar 25, 2007 7:00 am

sorenj07 wrote:Update! I'm probably going to use a 50VA, 9V toroidal power transformer as my OPT. The dual 115V primaries will act as a centertapped primary, and using 9V secondaries in parallel should yield me an impedance ratio of ~1306:8 ohms, which is perfect for the very low plate resistance of the 6S19P. I have no funds though so this is on the back burner :)


Did you cull this idea from a website in particular? If so, please link it. Interestings.

BTW, sorry your 6EM7 care package hasn't been sent yet - it will!
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby sorenj07 » Sun Mar 25, 2007 10:00 am

separks wrote:Did you cull this idea from a website in particular? If so, please link it. Interestings.

BTW, sorry your 6EM7 care package hasn't been sent yet - it will!


Well, no problem about the care package, any time is great :) As for the toroidal transformers, the idea has been around for some time to use them as PP outputs, due to toroids' pretty high bandwidth in general. One thread in particular kind of reminded me and made me realize that I could use one instead of finding a regular E-I transformer with such a low primary impedance. Here it is:

http://www.diyaudio.com/forums/showthre ... adid=98630

The mysterious, shadowy "sorenj07" figure on that forum might make you wonder..
User avatar
sorenj07
KT88
 
Posts: 925
Joined: Tue May 02, 2006 2:39 pm
Location: Berlin

Postby TomMcNally » Sun Mar 25, 2007 10:23 am

95 watts per channel ! woo hoo

Image
User avatar
TomMcNally
Darling du Jour
 
Posts: 2729
Joined: Sat Nov 19, 2005 2:19 pm
Location: Northfield, NJ

Next

Return to diy hifi

Who is online

Users browsing this forum: No registered users and 16 guests