Stereomour II 45 Conversion - Anyone Do It Yet?

[Deke609]

The manufacturer indicates that about 10V from the amp will be adequate for the LCD-4.  You won't get this from a #45 amp or a #45B amp unless you get custom wound output transformers with higher impedance taps to make more power available. 

I did some more reading this morning that i think clears this up. IN a FAQ, Audeze explains that the higher Watts recommendation is for transients. That makes sense to me since, according to a couple of online calculators I checked, the 97 dB/mV sensitivity of the LCD4 only requires 63 mW to get to 115 dB - which is pretty loud. 

Lundahl makes a headphone output transformer tapped for 4.6K/32, 5.4K/150 and 5.4K/600 ohms: LL2765. Assuming I can get it made for parafeed, do you foresee any problems with trying it?  Datasheet attached. 

I'd be curious to hear how the SII-45 performs driving the LCD4 with a ton more headroom.  The impedance/frequency plot of phones is nearly ruler flat, so I'm thinking the loss of damping with the secondaries wired for 150 ohms might not be a big deal.  But I suspect I'd need to pad down the signal, but not sure where in the BeePre -> SII-45 chain.

If it is workable, I could convert the SII-2A3 to run regular 45s with the LL2765, and compare it to the SII-45 rebuild for EML 45B.  I think that would be cool.

cheers and many thanks, Derek

 

[Deke609]

With 7W dissipated, that will be 75C rise, so you'll see a constant 100C during operation.  This is the absolute maximum dissipation that you could really run on these for very long, and putting the Crack over a laptop cooler to circulate air with fans will be a very good idea.

Hmmmm. Thanks for this PB. But I'm not sure I want to venture into the territory of potentially destructive testing! I'm not using the Crack but I'd like it to survive that so my son can use it down the line.

cheers, Derek

 

[Paul Birkeland]

I did some more reading this morning that i think clears this up. IN a FAQ, Audeze explains that the higher Watts recommendation is for transients. That makes sense to me since, according to a couple of online calculators I checked, the 97 dB/mV sensitivity of the LCD4 only requires 63 mW to get to 115 dB - which is pretty loud. 

You can play a 60Hz tone into your headphones that sounds "loud" and measure it with your AC volt meter and make this determination yourself.

Lundahl makes a headphone output transformer tapped for 4.6K/32, 5.4K/150 and 5.4K/600 ohms: LL2765. Assuming I can get it made for parafeed, do you foresee any problems with trying it?  Datasheet attached. 

Yes, noise.  The 32 ohm output might be quiet enough.  I'd also want to see frequency response measurements for each of those secondary configurations.

The impedance/frequency plot of phones is nearly ruler flat, so I'm thinking the loss of damping with the secondaries wired for 150 ohms might not be a big deal. 

Damping won't be a big deal, but noise will be.

I'm getting the impression that you should just use the Kaiju.  It has the power you need and doesn't require a tube that one random company makes. 

 

[Deke609]

Many thanks PB.

You can play a 60Hz tone into your headphones that sounds "loud" and measure it with your AC volt meter and make this determination yourself.

Yes, true. But I don't have a problem get loud enough. I'm just curious about whether more power for split-second transients is noticeable.

Yes, noise.  The 32 ohm output might be quiet enough.  I'd also want to see frequency response measurements for each of those secondary configurations.

Thanks, that's helpful. Does one component in particular contribute to the noise - e.g., is this heater/filament noise?

I'm getting the impression that you should just use the Kaiju.  It has the power you need and doesn't require a tube that one random company makes.

The Kaiju's definitely not going anywhere. And just be clear: I don't think there is anything wrong with my SII-45. This is project is mostly just for the fun/learning to be had: I want to see whether the SII-45 can be brought closer in performance to the Kaiju with my favorite 45 tubes, the EMLs (whether they are technically "45" tubes does'nt much concner me - they are drop ins for a regular 45 and to my ears sound better).

cheers and thanks, Derek

 

[Paul Birkeland]

Yes, true. But I don't have a problem get loud enough. I'm just curious about whether more power for split-second transients is noticeable.

If "loud" is 0.5V vs. 5V, doesn't that provide a ton of information for you about how critical more transient power would be?

Thanks, that's helpful. Does one component in particular contribute to the noise - e.g., is this heater/filament noise?

The #45.

 

[Deke609]

If "loud" is 0.5V vs. 5V, doesn't that provide a ton of information for you about how critical more transient power would be?

I'm just interested in testing Audeze's argument for having tons of power on tap: the brain interprets millisecond loud transients (e.g., 30-50 dB louder) that form the leading edge of notes/sounds as part of what makes things sound "realistic". Their argument can be found here: http://www.audeze.com/blogs/technology-and-innovation/sensitivity-impedance-and-amplifier-power I understand them to be suggesting that the transient edge, because of its super short duration, doesn't register as "loud" in the brain, but as part of what the brain expects as part of the sound of something real (drum hit, hand clap, etc.). I'm not saying Audeze is right - but that seems to be why they say tons of excess power is necessary.

The #45.

Ah. Thanks. So the noise in inherent to the tube structure/performance itself.

cheers, Derek

 

[Paul Joppa]

I'm just interested in testing Audeze's argument for having tons of power on tap: the brain interprets millisecond loud transients (e.g., 30-50 dB louder) that form the leading edge of notes/sounds as part of what makes things sound "realistic". ...

Lots of people say this. Recordings, however, virtually never capture more than 14-20dB of the transient peak. So OK, if you have your own microphones in Symphony Hall, and run your own wire from there to your living room, then this might be relevant.

 

[Deke609]

Lots of people say this. Recordings, however, virtually never capture more than 14-20dB of the transient peak. So OK, if you have your own microphones in Symphony Hall, and run your own wire from there to your living room, then this might be relevant.

Thanks PJ - On reflection, that makes immediate sense. I imagine that even with the microphones, capturing the full peak of the transients might be a challenge. I don't know anything about recording technology, but assuming mics make use of some sort of logarithmic translator of sound pressure to voltage signal, the receiving membrane would have to be insanely engineered to be able to deal with both low level and ultra-high level sound.

cheers and thanks, Derek

 

[Paul Joppa]

I looked up my notes - the OT-2 (Stereomour/SEX output transformer) operating as an 8K transformer (SEX) saturates about 2.4 watts at 25Hz, which is 4.8 watts at 35Hz or 9.6 watts at 50Hz. At 4K ohms (Stereomour), it's 2.4 watts at 18Hz, 4.8 watts at 25Hz, 9.6 watts at 35Hz, and 19.2 watts at 50Hz. Measured inductance was 45H at 350 Gauss, gradually rising to 265H at 7000 Gauss (as high as I could go at the time).

For comparison, OT-5 in Kaiju at 3K ohms is good for 8 watts at 22Hz.

These calculations are for 29-gauge M6 grain-oriented silicon steel at 13,500 Gauss flux density - moderately conservative since power transformers will sometimes push the M6 flux density to 16,000 Gauss.

A quick look at that Lundahl shows the primary is always the same, at all nominal impedances - only the secondary taps change. That's exactly the same as OT-2. Looks like it will handle about 10 times as much power as OT2 at any of the above impedance/frequency combinations. I see a rated 150 henries for the small gap (parafeed), but at an unspecified flux density. It's common to measure this at a high flux, but I don't know how Lundahl does it.

I estimate from the specs that Lundahl does their power rating at 16,000 Gauss, which is about 40% more power than 13,500 Gauss, but - again - I don't have specs on their core material, and I'm making some guesses.

 

[Deke609]

Many thanks PJ.

I see a rated 150 henries for the small gap (parafeed), but at an unspecified flux density. It's common to measure this at a high flux, but I don't know how Lundahl does it.

I estimate from the specs that Lundahl does their power rating at 16,000 Gauss, which is about 40% more power than 13,500 Gauss, but - again - I don't have specs on their core material, and I'm making some guesses.

I have emailed Lundahl and inquired about this. From what I read, going to amorphous core roughly doubles the inductance but also greatly increases the saturation and so reduces power handling - but with all that extra power handling available, I think this should be OK. Hopefully Lundahl will respond with answers.

cheers and thanks, Derek

 

[Deke609]

I've heard back from Per Lundahl. No specs for Gauss, and nothing specific for parafeed or amorphous core, but maybe this can be calculated/estimated from the info he provides? He states:

"The primary inductance for LL1679/PP(silicon iron core) is 150H  with 25u airgap, so the inductance for the AM core in PPZ (0u airgap)  should be at least 150H or higher. Measured at no DC component, 50Hz, approx. 100V RMS signal swing."

 

[Paul Birkeland]

So... Sowter 8983?

 

[Deke609]

And in case it is relevant, here's what Lundahl says (online) about the difference between silicon steel and amorphous core versions of its transformers:

"The obvious measurable difference between our silicon-iron cores and amorphous cores is that the saturation flux for the amorphous core is approximately 33% less than for the silicon-iron counterpart. This is caused partly by a lower saturating flux level, partly by a smaller fill-factor due to the thickness of the amorphous sheets. As a result, power bandwidth is reduced by about 50%. (This means that if the max output power for a standard LL1620/40mA is 25W at 30 Hz, corresponding max. power for LL1620AM/40mA is 13W.)"

 

[Deke609]

So... Sowter 8983?

I am clueless here. You guys left me in the dust as soon as the discussion went beyond my understandings that (a) other things being equal, bigger = more power, less saturation, and (b) amorphous may sound nice, like mumetal

I'm hoping PJ will jump in with more of his insight.

cheers and thanks, Derek

 

[Paul Joppa]

Yeah, that was a blast of information without enough context. Apologies.

I was basically trying to assess the operation of OT-2 at 8k ohms impedance and high power. Bottom line - to compare with Lundahl, one should use the same saturation flux density and frequency. Lundahl uses 30Hz, and I estimate from their specs that they use 16kGauss. Under those conditions, OT-2 can handle 4.85 watts.

 

[Deke609]

Yeah, that was a blast of information without enough context. Apologies.

No apologies necessary! After your post I did a little reading and thought I'd take a stab at estimating "Bmax" only to find that you need to know the # of turns and cross-sectional area of the core - neither of which I know, and can't see a way to estimate them based on what I do. So how you are able to make estimations is a total mystery to me! And no need to try to explain - even were I to apply myself rigorously to learning this stuff, I am likely a very long way from being able to understand (but making slow and steady "progress": I am getting more accurate in identifying what I don't understand)  ;D

cheers and thanks, Derek

 

[Tom-s]

There is no optimum capacitance. Bigger is better, limited only by practical considerations such as the start-up voltage sequencing, physical size of the cap, or high-frequency cap problems (generally worse the larger the cap). There is a minimum size, set mostly by the low frequency limit and the margin you choose. The analysis for a minimum size is pretty complicated because it involves the output circuit (plate choke, parafeed cap, OPT, and power supply) as well as the load (speaker impedance function). In practice, 47uF is good. I've seen lower values down to 20uF but I would not try it myself unless I could audition alternatives in the target system.

I've searched the forum and this thread in particular to find out how the cathode bypass cap of the 45 would need to change.
This is because i'm planning on using a 30uf or 60uf oil bypass in the 2A3 SII and might want to change to 45's later on.
How would going from 2A3 to 45 change this equation? How do i calculate this myself?

 

[Paul Joppa]

I've never done the full analysis, or set criteria which might be used with such an analysis to select a "minimum" cathode bypass capacitance. As in other cases, there is no optimum; larger is better but with diminishing returns - where do you draw the line?

That said, the 45 conversion goes lower in frequency than stock, so I would not go below the 47uF previously mentioned. Use the 60uF in preference to the 30uF.

 

[Doc B]

PJ's experience makes him pretty damned good at predictions. But at some point you have just dive in there, try some different cap values using his predictions as a starting point, take measurements and listen.

 

[Paul Birkeland]

I'll present my two lazy methods of looking at this...

#1.  Find a capacitor with a reactance that is 1/100th the value of the cathode resistor at 20Hz.  For a 1.6K cathode resistor, that's 50uF.  I don't have any analysis to back this up, but it's quick and dirty.   

#2.  Use this calculator.  Since we are interested in low frequency response, we can figure out the numbers for Rload and Rp based on the circuit itself.  A 40H choke (SMR wired for #45) has a reactance of about 5K at 20Hz, and the reflected load of the parallel feed output transformer is close to 4K.  The 5uF parallel feed capacitor will have an impedance of 1.6K at 20Hz, and this will be in series with the OT, so the Rload is 5600 ohms and the Rp is 5000 ohms.  I use 20Hz for the -0.1db box. For most of the other calculators on this website, there are formulas explaining how they created their calculators, but they are notably absent from this page.  When I put all these numbers in, that calculator spit out 44uF. 

I would imagine the Radiotron handbook has the appropriate formula in it, and that formula could be programmed into Excel to spit out cap values.  A part of me hopes that's what's going on with the MH-Audio calculator!