Welcome! Need support, you got it. Or share your ideas and experiences.
Return to MagneQuest/Peerless Forum
In Reply to: Re: MQ introduces a new limited series of output tranneys posted by Kyle K on May 15, 2004 at 10:48:45:
Hello Kyle:you wrote;
:::You said you didn't intend for these to be "universal" OPT's, but could one get reasonable performance using an 8 ohm load on the 16 ohm tap? Say, for a 2A3 application where you will not need the full power capability of the transformer?::::
let's see where this goes... here is how I would evaluate it....your impedance ratio from the 16 ohm tap to the full primary is 312.5:1 which means that if you use this tap then the reflected impedance to the primary is 312.5 times the value of impedance hooked to that tap.
your idea/proposal is to use an eight ohm speaker on the sixteen ohm tap... if we do this.. then the reflected primary impedance will be 2500 ohms.... which is what a 2A3 frequently likes to see as a primary impedance.
Let us assume your using the RH-60... most 2A3's are run at 60 mils of plate current. so far so good.
firtst thing I notice is that our dcr is going to be a bit high.... the dcr of the RH-60 is 180 ohms. Relative to a 5K primary 180 ohms is quite respectable... about a .4 db insertion loss or thereabouts... but for a 2.5K primary the copper resistance is on the high side.... so our insertion losses will be greater... probably not the end of the world...inductance wise.... we are still energizing the same number of turns on the same size core with the same size air gap... so we will have 30 henries of primary L available.
Now... ratio'ing down.... this actaully sort of helps us... 30 henries at 30 hertz will have an inductive reactance of 5655 ohms... which is more than 2 times the 2500 ohm impedance reflected to the primary... so we will gain bottom end extension (i.e., we will get better bass)....
if we calculate the phase angle of our load impedance... ratio'ing down to 2500 ohms will actually give us a moreso non-reactive effective impedance.... see our website for a discussion of phase angles as they relate to primary impedances.... but... make a long story short... we aren't getting hurt here...Now let's see what our power delivery capacities and capabilities are. Suppose our output power from the single 2A3 is 3.5 watts. Across 2500 ohms. therefore, our AC signal volts at full power will be 93.54 vrms.
Now... our airgap hasn't changed... it's fixed... but our ac flux would change. If we kept the same low frequency cut-off point as published for the RH-60 (45hz) then we would operate at considerably lower flux density then the design center of the RH-60 since it was rated for 10 watts output at 45 hertz.
skipping through a few computer iterations in preparing an answer for you... allow me to get to the punch line....
with a 2500 ohm primary and 3.5 watt output and 60 mils of dc current... you would be able to deliver full power down to 20 hertz before hitting our pre-established maximum flux density limit... so, again, your in pretty good shape....
so...now we know the core is robust enough electrically to handle your intended full power down to 20 hertz in this modified application....
but... do we have sufficient inductance to go that low.... the -3db power low frequency limit is that point or juncture wherein the inductive reactance equals the reflected impedance. 30 henries at 14 hertz has an inductive reactance of 2639 ohms which is just a little higher than 2500 ohms. the inductive reactance of 30 henries at 143 hertz is about 2450 ohms... so we would say that our -3db point will be at 14 hertz....
so... in this case we will run out of magnetic headroom before we run out of "inductance"... in-other-words the allowable flux density limits us to delivering 3.5 watts at 20 hertz.
Not so bad... heh?
As a further study.... let's see if we get into trouble with temperature rise either in the copper circuit or in the core....the RH-60 if operated at it's published design center... has a coil temp rise of approx 16 degrees C and a 3 degree C rise in the iron....
your proposed modified operation of the RH-60... will result in a coil temperature rise of 13 degrees C and an iron temperature rise of 1 degree C.
so.... your not going to run into any problems here either.... but this is some of the stuff your should check in making these kinds of decisions...
Conclusion... unless I've missed something.... I stand corrected in my dire, dire warnings... the only penalty we've been able to identify in your proposed use/application is that your insertion losses are going to be a bit higher... and I'm too lazy to calculate this out.... but as an estimate I would say that your going to have around a .65 ot .7 (point six five to point seven) db insertion loss... if you can live with this.... it's a bit higher than I like to design at... you should be fine....
I hope I've been helpful....
but... recall... this is ratio'ing down NOT UP... if you wanted to ratio up... it would not have worked out nearly so well...
moral of the story is... ratio'ing down generally only penalizes you in respect of increasing your copper losses.... ratio'ing up has many more penalties due to the finite inductances of the transformer and the greater magnitude of signal volts required for a given power level... so that the core will get taxed moreso... but.... that's a different study... just trust me on this :=)
MSL
Follow Ups:
Depends on how much interleaving of course. The transformer will have both leakage inductance and capacitance. Together these will resonate at some (usually ultrasonic) frequency. By cutting the load impedance in half, you will reduce the "Q" of that resonance. If it's already overdamped, then you would lose treble; if it's underdamped you might just get less ringing. In my subjective experience, underdamped sounds better as long as you don't lose more than a few dB at 20kHz.Here's a thought - with an active crossover, run a 45 for the treble at 5k, and a 2A3 or 300B at 2.5k for the bass.
Hi Paul:Within minutes of hitting the send button... I was already thinking of writing an addendum pointing toward the consideration you mention above re: leakage inductances, cacpacitances,and etc.
I had thought of pointing out that ratio'ing down... though much more preferable (i.e., not as harmful as ratio'ing up in most cases)... still... none-the-less is not a substitute for the proper part for the job....
so... that... say a 2.5K "Friar Tuck" trans... built along the same performance profile as the 5K Robin Hood... would look quite different electrically and be moreso optimized for 2.5K as opposed to a ratio'ed down RH. I agree.
the 2.5K trans would have fewer turns, thinner pads btwn windings, and the net leakage inductance would be half of the RH at 5K. And since the AC voltage gradients are different... the insulation package would be different and optimized for the 2.5K impedance.
So... yep.... "universals" aren't optimized fully for each condition and that is an argument I've made for years and basically why I still prefer to recommend that folks use transformers at their design centers.... i.e., use them where they have been optimized.
now...imagine a 5K three way universal.... rig it for 2.5K, 5K and 10K... and then study it's performance... do FR tests at each impedance, calculate insertion losses, do square waves to see how each tap will behave differently, do low frequency power response plots for each.... and you would find that the unit behaves best at it's nominal design center.
"Universals" (one size fits the skinny, medium and fat man) are great for manufacturer's because they can offer or cover three or more different applications from one transformer... but... does each of these "impedances" have equal performance for each tap or operating point? Nope.
In the case of ratio'ing down (as the poster orig inquiried about)... far less harm is generally imposed on the performance.... not ideal by any means... because your moving away (by defintion) the ideal (optimized) point of operation.... by less harm than ratio'ing up...
and after studying the original posters idea... I was a bit surprised, that the RH actually moved perhaps less off of "ideal" than I might have suspected... hence, my willingness, to provisionally back off of my *dire* warnings.
if the Robin Hood series goes well perhaps we should follow it up with a Friar Tuck series for 2.5K or 3K. For there still remains no equal substitute for using a properly designed part at it's design center for the job you need to do...
MSL
Still, though, for seventy-five bucks each ya gotta admit it's an attractive and versatile option compared to a lot of other alternatives out there.
.
This post is made possible by the generous support of people like you and our sponsors: