4:1 vs 3:1 output freq response

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

BluegrassDan

Well-known member
GDIY Supporter
Joined
Mar 17, 2009
Messages
507
Location
Elizabethton, TN
Tweakin' and attemptin' and learnin'...

Here are sweeps of increasing gain comparing a 4:1 Jensen output with a 3:1 Peluso (Tamura) output with 600 ohm (actually 650) termination. A few questions:

1. Is the decrease in frequency response indicative of the ratio difference? E.g., The circuit can't drive the 3:1 transformer into 600 as well as the 4:1 at higher gain? The high and low frequency rolloff gets worse when the Peluso is strapped for 2:1 and 1:1.

2. Is there a way to modify the 12BH7 WCF to give more drive to the 3:1? Or, is there a better option in this position? SRPP? ECC99?  Something else?
 

Attachments

  • Jensen 4 to 1, cathode bypass test.jpg
    Jensen 4 to 1, cathode bypass test.jpg
    676.3 KB · Views: 58
I don't see a problem. They both look great to me. The 3:1 range is 25Hz - 20kHz.

If you really don't want to stress the transformer at all with 3:1, just increase the 650 to 1K.

But I would really listen to the 3:1 with 650. If you believe that transformers can sound different, loading it on the edge is probably the best (only?) way to make a transformer sound different (LF distortion).
 
CJ said:
can you run an A/B  without the circuit and un-terminated secondary?  maybe at 10 volts input each?

I sure can, soon as I get home.

Also should mention that these sweeps were done with what I think is a pretty high value coupling cap, 47uF. I can get away with 4.7uF on the 4:1, with a bump around 10Hz, but the 3:1 has a bigger bump and steeper rolloff.
 
They all look flat down to 20Hz to me. The high frequency changes with gain are more likely to be to do with the open loop gain of the previous stages than the output stage itself. With Cinemag transformer you would expect this to extend easily to 40K.

You can also see the effects of open loop gain in the sub 20Hz region.

Cheers

Ian
 
Sweep the amplifier without the output transformer so you have a baseline.  Use a 10K load for this.  I  suspect that you'll  still see  a bit of a drop off at the high end due to the Budweiser effect.  Or Miller.  Whichever you prefer.
Some of this could be mitigated by adding a follower after V1a and using a lower value level pot. 
Your 12BH7 is plenty good to drive a 3:1 transformer so the low end deficiencies with the Peluso are probably just inherent with that particular transformer. 











 
BluegrassDan said:
1. Is the decrease in frequency response indicative of the ratio difference? E.g., The circuit can't drive the 3:1 transformer into 600 as well as the 4:1 at higher gain? The high and low frequency rolloff gets worse when the Peluso is strapped for 2:1 and 1:1.
You will be sure that  the WCF can't drive the load if rolloffs are present also when the Jensen is used with lower load. 
Jensen 4:1 with Rl=360 vs. Peluso 3:1 with Rl=650.
What's the output voltage in volts at 80dB? Have you monitored  THD during the measurement?
Here https://www.pearl-hifi.com/06_Lit_Archive/02_PEARL_Arch/Vol_04/Sec_19/991_White_Follower_Optimization.pdf
you can find lot of info about WCF optimisation.

Btw, may I ask which software and hardware you are using for measurement?

 
Okay...here are some sweeps and observations...

First, this graph is "calibrated" to the gain settings on the preamp. At 55dB of gain, the preamp is spitting out 10 volts rms into the output TX.

Second, I suspect the high end rolloff is mostly due to the Metric Halo Lio8 sound card. My oscilloscope does not indicate this rolloff. It does, however, indicate the low frequency rolloff conundrums.

These sweeps are of the Jensen loaded with 650ohms versus the sound card @ 10k. 35dB and 55dB gain.

(Disregarding the high frequency stuff, this one seems pretty good.)
 

Attachments

  • Jensen 4 to 1, load vs no load.jpg
    Jensen 4 to 1, load vs no load.jpg
    545.8 KB · Views: 21
Next, here is the Peluso 3:1 loaded at 650ohm vs 10k. I dialed the gain back to 33dB and 53dB, as to not overload the soundcard.

The load difference seems to change this frequency response very little.

According to John Peluso's spec sheet, "The transformer is DC gapped and has 4 identical 150 ohm windings that are trifilar wound. So can use a 1:1:1:1 or 1:2, 2:1, 3:1 applications."

He also explained to me (and I don't understand) that this transformer is not a "floating" type. That (if I understand at all correctly) the circuit is actually "seeing" 150 ohms, not just reflecting what is on the secondary. (Again, I could have heard and understood this entirely wrong.)

But it leads me to wonder. Any thoughts?
 

Attachments

  • Peluso 3 to 1, load vs no load.jpg
    Peluso 3 to 1, load vs no load.jpg
    570.2 KB · Views: 13
Finally, here is the Jensen 4:1 loaded at 360 ohm instead of 650.

Other than a slight drop in signal, the frequency response is the same. This leads me to believe that the 12BH7 WCF has plenty of "oomph" to drive the 3:1.

I measured the 12BH7 output impedance to be 107 ohms. Seems like it should be able to drive up to 2:1 or maybe 1:1.

My WHOLE goal in this is to use a 3:1, 2:1, or 1:1 output TX for more output gain and less saturation on the preamp. I also want to try a lower 1:5 input TX, and switchable 1:10 or 1:20 for ribbon mics, if I can get the output level up more.
 

Attachments

  • Jensen 4 to 1, 360ohm.jpg
    Jensen 4 to 1, 360ohm.jpg
    509.2 KB · Views: 8
Ok. But is it a microphone output transformer or a proper high level line output transfomer? If it's significantly smaller physically, that will pretty much equate to less bandwidth / power and your results are as expected.
 
can you go up to 500 K Hz with zero load and compare the two transformers?

maybe limit graph from 20 KC to 500 KC?  thanks!

everybody is flat from 20 to 20, for the most part, there are some oddballs, like the K-241-d with the rise at 12 K,

so the finger print of the transformer lies in the self rez points usually starting around 50 K and higher,

these peaks are related to leakage inductance and capacitance, hopefully scope input z is 1 meg or better,



 

Latest posts

Back
Top