[RollE2k]

Good to hear, i have bought a pair of used ATH-w1000x which are 42ohms that i will need a new headphone-amp for soon, so this looks as a nice project!
Depending on the step-down value of the transformers i was thinking about trying top-mounted tubes 6n1p or ECC88's. since i have some nos of these.
Will check the pictures and waiting answer regarding the step-down value!

 

[dsavitsk]

Voltage gain into 32 Ohms is ~2x. Into 300 Ohms it is about 6x.

 

[RollE2k]

Quote:

Voltage gain into 32 Ohms is ~2x. Into 300 Ohms it is about 6x.

Great to know, then there is probably possibility to put lower-gain tubes there also. Do you have exact step-down ratio or is that a secret?
Also when running tubes without common cathode, it is not possible to set current separately for each tube half right (i'm thinking to set more exact current between both triode-halves) it doesn't look like that from the schematics at least.

 

[yuujin]

hi guys, i know the bijou is different in design, but i would like to ask if anyone has listened to both of them?

 

[dsavitsk]

Great to know, then there is probably possibility to put lower-gain tubes there also. Do you have exact step-down ratio or is that a secret?

Also when running tubes without common cathode, it is not possible to set current separately for each tube half right (i'm thinking to set more exact current between both triode-halves) it doesn't look like that from the schematics at least.

I am not aware of any lower mu tubes that will fit the board.

mu is 38. The transformer is 10K:32 or 10K:300. That means that the turns (step down) ratio is ~17.5 for the low Z setting and 5.8 for high. This gets you slightly higher gain than I suggested above, but if you account for copper and core losses, etc, it is about 2 and 6.

The cathodes are tied internally, so there is no way on this amp to set current independently. I suppose you could in theory use resistors on each plate to try to equalize things, but there is no reason to do so.

 

[RollE2k]

Quote:

I am not aware of any lower mu tubes that will fit the board.
mu is 38. The transformer is 10K:32 or 10K:300. That means that the turns (step down) ratio is ~17.5 for the low Z setting and 5.8 for high. This gets you slightly higher gain than I suggested above, but if you account for copper and core losses, etc, it is about 2 and 6.
The cathodes are tied internally, so there is no way on this amp to set current independently. I suppose you could in theory use resistors on each plate to try to equalize things, but there is no reason to do so.

I was thinking about maybe mount the tubes at the top of the case, so 9-pin socket would fit then. But still, its probably better getting matched/balanced tubes then.
Last questions then; 
Approx what current are the tubes running, and how much can you "up" the default CCS?
And one thing i don't get about parafeed - when running a transformer as a anode load (like normal SET) the transformer does simply emulate a impedance loading for the tube, and its really important to match with right step-down ratio. But when running parafeed isn't that as critical - that means the more step-down the better (at least as long as the gain is enough). Am i right about this?
If that above is true then lowering plate resistance (with another tube for example) would mean that you could easily drive lower impedance headphones.

 

[dsavitsk]

If you use a different tube with the socket mounted off board, then yes you could match the current. But, I really think it is unnecessary.

The tubes run at ~15mA. I would not increase it too much as we are running the power transformer at close to the limit. This is particularly true w/r/t/ heater current, so don't use a tube that will increase that.

In parafeed, the transformer is in parallel with the CCS and presents the "real" load. It is incorrect (though repeated all over the web) that there is an infinite load on the tube.

Driving headphones is not like driving speakers. You do not need a super low Z or a high damping factor. The back EMF is much lower, the load lines are nearly flat, and there are generally no crossovers to worry about. So, a super low output impedance is just not that important. This amp will drive just about anything you throw at it.

This might help with your other questions, and it explains the parallel load issue in more detail: http://diy.ecpaudio.com/p/parafeed-tutorial.html

 

[tomb]

Just curious, but would anyone be interested in bronze-brown (root beer?) anodized cases or just black?  I'm talking about something similar to the Torpedo website color scheme.

 

[Chet_Summers]

If you do black I'd have to break down and order a short kit.

 

[scootsit]

Just curious, but would anyone be interested in bronze-brown (root beer?) anodized cases or just black?  I'm talking about something similar to the Torpedo website color scheme.

My opinion shouldn't really count, because I'm way too happy with my Millett to be wanting. BUT, I'll offer it anyway.

I think you should go for a cream yellow like the old heathkits. Something super vintage looking. Maybe if it had the hammered texture to the finish with the brown, which was not unheard of on vintage amps.

 

[KimLaroux]

Quote:

Finished up yesterday:
 

 
[...]

 
I think this design should be made the final kit. It's simply outstanding. The color scheme, the textures, the logo plate, the hardware... it fits perfectly. It has more charm than the plain aluminum kit. The plain aluminum enclosure reminds me of industrial power supplies. =P Not something that would blend in an hi-fi setup, anyways.
 
Steampunk winz. The only thing I would modify on this build is pull the tubes about 1 cm lower, so that the sockets don't show.

 

[tomb]

Hmm ... I can't do all of that, even though I agree.  What if the case were black and the endplates anodized bronze in the pic?

 

[UKToecutter]

I wonder if someone can assist me with some strange test results I'm getting.
I've been testing the tubes that Tom supplied me with.
6J6's and 5964's give me results I would expect.  The 5844's however give me some very strange results.
 
As an example, here is the test results of a 5964;
 
 

Test Results		[size=medium]Tube[/size]		[size=medium]Method[/size]
		[size=medium]5964[/size]		[size=medium]Fixed Bias[/size]
Item	[size=medium]Unit[/size]	[size=medium]Spec [/size]	[size=medium]Result 1 [/size]	[size=medium]Result 2[/size]
Plate Volts	[size=medium] Ep vdc[/size]	[size=medium]100[/size]	[size=medium]100[/size]	[size=medium]100[/size]
DC Plate Current	[size=medium] Ip mAdc[/size]	[size=medium]9.5[/size]	[size=medium]7[/size]	[size=medium]6.8[/size]
AC Plate Current	[size=medium] Ip mAac[/size]	[size=medium]~[/size]	[size=medium]0.531[/size]	[size=medium]0.485[/size]
Screen Volts	[size=medium]G2 vdc[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Screen Current	[size=medium]IG2 mA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Bias Volts	[size=medium]G1 vdc[/size]	[size=medium]1[/size]	[size=medium]1[/size]	[size=medium]1[/size]
Test Signal Volts	[size=medium]RMS[/size]	[size=medium]~[/size]	[size=medium]0.1[/size]	[size=medium]0.1[/size]
Mutual Conductance	[size=medium]uMhos[/size]	[size=medium]6000[/size]	[size=medium]5310[/size]	[size=medium]4850[/size]
Grid #1 Leakage	[size=medium]uA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0.01[/size]
H/K Leakage	[size=medium]mA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Plate Resistance	[size=medium]RP ohms[/size]	[size=medium]~[/size]	[size=medium]5.5K[/size]	[size=medium]7.1K[/size]
Tube Gain	[size=medium]Mu[/size]	[size=medium]~[/size]	[size=medium]29.2[/size]	[size=medium]34.4[/size]
Heater	[size=medium]eF vdc[/size]	[size=medium]6.3[/size]	[size=medium]6.3[/size]	[size=medium]6.3[/size]
Heater Run Time	[size=medium]Sec[/size]		[size=medium]65[/size]	[size=medium]86[/size]

 
So all looks reasonable.
 
I pulled the 5844 datasheet and the typical operating parameters are Ep vdc 100V and Ip mAdc 4.8mA with uMhos 3700
I set my tester to these characteristics and ran the test.
Results were;
 
 

Test Results		[size=medium]Tube[/size]		[size=medium]Method[/size]
		[size=medium]5844[/size]		[size=medium]Fixed Bias[/size]
Item	[size=medium]Unit[/size]	[size=medium]Spec [/size]	[size=medium]Result 1 [/size]	[size=medium]Result 2[/size]
Plate Volts	[size=medium] Ep vdc[/size]	[size=medium]100[/size]	[size=medium]100[/size]	[size=medium]100[/size]
DC Plate Current	[size=medium] Ip mAdc[/size]	[size=medium]4.8[/size]	[size=medium]10.1[/size]	[size=medium]10.1[/size]
AC Plate Current	[size=medium] Ip mAac[/size]	[size=medium]~[/size]	[size=medium]0.6[/size]	[size=medium]0.61[/size]
Screen Volts	[size=medium]G2 vdc[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Screen Current	[size=medium]IG2 mA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Bias Volts	[size=medium]G1 vdc[/size]	[size=medium]1[/size]	[size=medium]1[/size]	[size=medium]1[/size]
Test Signal Volts	[size=medium]RMS[/size]	[size=medium]~[/size]	[size=medium]0.1[/size]	[size=medium]0.1[/size]
Mutual Conductance	[size=medium]uMhos[/size]	[size=medium]3700[/size]	[size=medium]6000[/size]	[size=medium]6100[/size]
Grid #1 Leakage	[size=medium]uA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
H/K Leakage	[size=medium]mA[/size]	[size=medium]0[/size]	[size=medium]0[/size]	[size=medium]0[/size]
Plate Resistance	[size=medium]RP ohms[/size]	[size=medium]~[/size]	[size=medium]10.3K[/size]	[size=medium]11.1K[/size]
Tube Gain	[size=medium]Mu[/size]	[size=medium]~[/size]	[size=medium]61.8[/size]	[size=medium]67.7[/size]
Heater	[size=medium]eF vdc[/size]	[size=medium]6.3[/size]	[size=medium]6.3[/size]	[size=medium]6.3[/size]
Heater Run Time	[size=medium]Sec[/size]		[size=medium]215[/size]	[size=medium]236[/size]

 
It looks to me as thought the Mutual Conductance and DC Plate Current are wrong.
 
Have I got a rougue 5844 datasheet??
 
Any assistance appreciated.
 
Andy
 
 

 

[tomb]

I'll study your figures more when I get some time, but just an FYI - the Torpedo runs at 200V on the plates, whereas the maximum plate voltage on 5844's is 175V.  So, the 5844's are being over-driven in the Torpedo.

 

[UKToecutter]

Strangely enough the GE datasheet does specify a maximum anode voltage of 200V