Andy
I am using LL1664, cheaper than1620. Don't know if you read my post from last night where I posted my PSE amp
Radu
Yes - your amp looks great. Did you ever compare the LL1664 to the 1620 or 1623? I directly compared the LL1692 to the LL1660 and I thought the LL1660 was slightly clearer. I think Thomas Mayer preferred the LL1692.
Very nice construction - extremely impressive.
How are you finding the paper cone Alpairs? I have the metal cone Alpair 10s, and also four of the CHP70P which are still to be used in something.
Last edited:
Andy,
Never compared 1620 to 1664. Initialy I tried UBT-1 that I had on hand and I really liked the sound. So I went for a less expensive Lundahl( One electron ubt-1 didnt have enough Lp plus it is 1.6k/4/8/16). I had struggled to choose the driver, but I didn' t have any other dht or idht option that sounded better. i like both 1660 and 1692A sound. I went for 1692A to allow more driving capability, and Alt T fitted my design needs.
I like the Alpair very much. They sound very even on all frequencies. I have only played them for 20 hours, so. I am waiting to brake em in.
Thanks for your kind words about my amp. This is the secon with 4P1L. First is a parallel PP that uses 8 of them and the pse 6 of them. I am working now on a SE head phone amp, the third in this series with 4 P 1 L. I need big reserve of this great tube.
Best,
Radu
Never compared 1620 to 1664. Initialy I tried UBT-1 that I had on hand and I really liked the sound. So I went for a less expensive Lundahl( One electron ubt-1 didnt have enough Lp plus it is 1.6k/4/8/16). I had struggled to choose the driver, but I didn' t have any other dht or idht option that sounded better. i like both 1660 and 1692A sound. I went for 1692A to allow more driving capability, and Alt T fitted my design needs.
I like the Alpair very much. They sound very even on all frequencies. I have only played them for 20 hours, so. I am waiting to brake em in.
Thanks for your kind words about my amp. This is the secon with 4P1L. First is a parallel PP that uses 8 of them and the pse 6 of them. I am working now on a SE head phone amp, the third in this series with 4 P 1 L. I need big reserve of this great tube.
Best,
Radu
That day belongs to 2011I already posted a diy version that has the same or better performance. I used Rod's initial idea, a cap multiplier (or gyrator) in front of the filament, a CCS after the filament (credit to Rod for the idea), and created a circuit up to the performance I wanted. You need to know what you're doing to adjust some of the circuit's resistor values to set the current. This is not a copy of Rod's commercial circuit, which has never been published, as far as I know.
P.S. Rod's initial circuit that I started with can be found here http://www.diyaudio.com/forums/tubes-valves/38248-new-dht-heater-7.html#post2024952
Can you share QD44H11 datasheet or tell an equivalent transistor? Uncle Google tells almost nothing about it? Is there any particular reason you have used them (except having them by hand)?
Google "44H11". There are various prefixes to many devices depending on the manufacturer. It's a NPN transistor. Same pinout as the type of devices Rod uses. I leave it to them to explain their choices - there are quite a few possible alternatives, some of which have been discussed on threads about filament boards.
Last edited:
The LL1692 would allow one configuration that can be rarely used with IT coupled amps. The ALT Q connection would allow true Class A2 drive without issues. Biasing the output 4P1L's at about 250/30 mA each for about -23.5V fixed bias one could think to get a max Pout of 5W with 30V peak drive and 2.5K load.
Using the LL1692 in ALT Q with 3.5:1 step-down ratio would mean the driver should be able to provide 105V peak. This is still pretty achievable thanks to 95H @ 21 mA of the primary. Using another 4P1L at 20 mA with enough plate voltage would do the trick. What you get at the output of the LL1692 is 30V peak but with the ability to pump into the grids about 60 mA peak current and the output impedance of the 4P1L driver would be in the region of 120R only! The DC resistance of the LL1692 secondary would only be around 50R which is 10 times lower than the typical recommended limit of 500R. Last but not least the capacitive load would be about that of a single 4P1L!! Pretty good I would say. Another good choice would be the 6H30 that has more gain and Radu has already used successfully for the PP. Actually it could also be used with the two sections in parallel for even lower Zout....
In my opinion the LL1627 is the best choice because can be used for minimal loss (0.25 dB power loss). The LL1664 or the LL1623 at 3.3K and 3K connectio nare second best with 0.5 dB power loss. THe LL1664 has the advantage of the lower price at least.
The LL1620 is used at 0.8 dB power loss. Usable but not ideal. It's a waste especially considering the already low Pout of the amp. If you want to try the LL1627 I have a pair and you could borrow them.
Last edited:
No Andy you are not using that connection. The LL1620 will have 0.25 dB power loss if you use it for 11.5K primary impedance. For 3K it is 0.8 dB because the contribution to impedance from the primary and secondary DC resistance increases.
In the first case you have connection B with 38.4 ratio for 11.8K and DC resistances are 308R and 0.2R (which translates to 295R at the primary). From this you have that efficiency is better than 95% (i.e. about 0.22 dB).
For 3K the ratio will be 19.2 and the secondary Rdc=0.8R. The secondary Rdc is 0.8R x 19.2 x 19.2 =295R and the primary DC resistance stays the same (from this you already know without calculations that its impact on efficiency will be more) for a total of 603R. Efficiency is 83% (i.e. 0.8 dB power loss.
The power handling increases to 50W but you don't need it here.
Copper loss is just 1 of many design factors. If two transformers have the same frequency respons the transformer with the lowest copper loss will be the best. (sounds more open, clean, best soundstage).
For the low frequencies you need a large core and much windings, copper loss can be high in some designs and without propper winding techniques the high frequency response could be poor too. Not forget resonance problems.
A good design is very difficult. Tango did all* very good, that is why they are famous.
* not all Tango but some high end series they made.
For the low frequencies you need a large core and much windings, copper loss can be high in some designs and without propper winding techniques the high frequency response could be poor too. Not forget resonance problems.
A good design is very difficult. Tango did all* very good, that is why they are famous.
* not all Tango but some high end series they made.
Thanks! But Tango has stopped production. So who now is making what you see as the best designs - Hashimoto, Tamura.......?
Tamura and Hashimoto makes some nice transformers, Monolith Magnetics too.
Monolith Magnetic is the only one who has excellent information about there transformers.
Monolith Magnetic is the only one who has excellent information about there transformers.
I looked through the Monolith website, they too are offering double C-core nanocrystaline cores, not toroids.
I have been looking through this thread (not finished reading it fully, I admit) and am seriously thinking about a build. I was originally considering Thomas Mayer's 6bc5a amp. I like the low cost of the output tubes. An all 4p1l PSE amp is a bit more complicated and probably more expensive but I like the idea of an all direct-heated amplifier.
Here is my question(s): I would like the amp to be one box. How much would the sound degrade to simplify the power supply just a bit? Would it be possible to get away with one filament supply for each channel? (and thus have only 3 power transformers on board--save space and keep the EMF down.)
Also, er, ehem... I don't really have a preamp right now. I have a Nelson Pass B1 buffer but have been shying away from that lately. Would a 4p1l driver stage be sufficient for an decent line input with a volume pot at the input? My sound card and phono stage seem to have a decent amount of oomph. They sound fine straight into the Simple SE I am currently using. (Better than with the B1 in the way, at any rate.)
Any input is very much appreciated.
Thanks!
John-
Here is my question(s): I would like the amp to be one box. How much would the sound degrade to simplify the power supply just a bit? Would it be possible to get away with one filament supply for each channel? (and thus have only 3 power transformers on board--save space and keep the EMF down.)
Also, er, ehem... I don't really have a preamp right now. I have a Nelson Pass B1 buffer but have been shying away from that lately. Would a 4p1l driver stage be sufficient for an decent line input with a volume pot at the input? My sound card and phono stage seem to have a decent amount of oomph. They sound fine straight into the Simple SE I am currently using. (Better than with the B1 in the way, at any rate.)
Any input is very much appreciated.
Thanks!
John-
"I would like it to be in one box" - sure, just a bigger box if you like one box. Keep the PSU parts all on one side of the OPTs and the filament boards+heatsinks, the tubes and the cathode resistors (they get hot) on the other half of the chassis. i.e. keep the magnetic components right over on one side and the filament boards and the input tubes the other end. Instead of using an interstage you can use an active load and a capacitor, as Ale has outlined on his Bartola website. That will probably save money and space. I'd favour a Russian teflon FT-2 but it's a personal choice.
"One filament supply per channel" - you may still thinking of the filaments in filament bias like heaters. The signal goes THROUGH the filament supply. If you simplify it you will hear it. You can skip the choke in the filament supply, but at least use a good EI transformer for EACH filament supply - they all have to be discreet and separate. Four filament supplies. Also remember the heatsinks take a little space. You could put them on the back of the chassis.
The signal PSU is up to your taste. I think a regulated first stage sounds cleaner so I'd keep that in. There's a Salas reg, or use a VR tube preceded by a simple DN2540.
If you have 2v in, then you should be OK with average sensitivity speakers (89db) in a smaller room and not at high sound levels.
You should like the sound you get!
Last edited:
Hmmm... So what would I need in Iron, exactly, to start this?
4 Split bobbin PTs for the Rod Coleman (or similar) regs (should these get a CRC or CLC for the raw supply? I am imagining these would be solid state rectified, no?) What should the current rating be on these?
1 PT for the HV around 200ma? If using a Salas shunt reg, what would the ideal secondaries be? Probably do a Tube rectifier for the HV? then CLC or LCLC filters?
2 IT tranformers
2 OPTs.
Alternatively, does anyone (Andy?) have a schematic of the power supply they used?
Thanks for the help!
John-
4 Split bobbin PTs for the Rod Coleman (or similar) regs (should these get a CRC or CLC for the raw supply? I am imagining these would be solid state rectified, no?) What should the current rating be on these?
1 PT for the HV around 200ma? If using a Salas shunt reg, what would the ideal secondaries be? Probably do a Tube rectifier for the HV? then CLC or LCLC filters?
2 IT tranformers
2 OPTs.
Alternatively, does anyone (Andy?) have a schematic of the power supply they used?
Thanks for the help!
John-
If you're following my schematic, you'll need:
each PSE 4P1L - 28v DC supply at 1.25A, solid state rectification, 20,000uF caps approx., Hammond 159ZC choke if you want extra smoothing. Can be choke input or small cap input, like up to 470uF, which can fine tune the voltage out.
each 4P1L - 18v DC supply at 620mA as above, with Hammond 159ZA choke.
PSU - you'll need about 275vDC at 200mA, so allow a margin above 200mA. You can use a typical mains transformer with extra 6.3v winding, and then rectify with two damper diodes like 6DT4 or similar. Or you can use two 6BY5G in parallel (twin damper diodes) which sound good and have a little extra voltage drop. Choke input is nice, or use a small cap like 1uF. Up to you to work this out. I use only motor runs in the PSU, no electrolytics. A second choke would be a good idea. If you're ordering from Hammond they have plenty of this stuff, or you may have parts from other sources.
All this adds up, which is why I use a separate PSU chassis. It also allows me to experiment with filament supplies, like adding chokes etc. I use a 4-way Neutrik Speakon for the HT and earth and a couple of 4-way XLRs for the filaments. You could use one 8-way Speakon, though I find them fiddly to wire up. When I was developing this I used three PSU chassis - the HT, and two separate filament supplies in cheap black plastic cases. I played around with different options to see if I could hear improvements. Once you start developing this you want to optimise everything, and I don't like to be restricted for space from the start.
As for OPTs, I've been searching for the best choices. At present I'm using a LL1660/80mA which is pretty good but I'm open to other ideas. Options could include James 6123HS, Daburu DA-35X, Silk S-330F, Softone RW-20, Hashimoto H-20-3.5, Monolith Etude 1, Tamura F-7002. Basically 3k to 3.5k, 80mA gap, about 3 kilos weight and 20 watts to get decent bass - whatever you'd use for a 300b. The smaller 1.6 kilo 12 watts types probably also sound good but with lighter bass. I don't know the US transformer makers, I'm in Europe.
For interstage I've been quite happy with LL1660. I use the 18mA version because I had a pair, but 25mA would be better. The 4P1L sounds progressively fuller and more dynamic above 15mA, which is as low as you'd go (here it sounds thinner and rather anaemic). 18mA works but more is better. The Hammond 126C (or probably 126B but I didn't try it) works and actually sounds quite good, but I'd want better. It's worth optimising this design. Again, you could look at Hashimoto, James and the rest of the bunch.
Since the OPT is identical to what you'd use for a 300b, and so is Rod's regulator (though not the filament supply which is higher voltage for filament bias), and since a 20mA interstage is quite generic, some existing 300b amps could be converted for PSE 4P1L. I started with a 300b SET myself, which had 4P1L driver and a 26 preamp. Going to 2 stages of 4P1L in filament bias was quite noticeably cleaner.
Last edited: