Poor low frequency response <100 Hz

sweet & juicy SE amp for 1626 Darlings and the 6L6 family

Poor low frequency response <100 Hz

Postby MKJ » Sat Nov 10, 2012 6:20 am

I've built a Darling that is very similar to the Clementine. That might sound a little bit academic, but I've never liked the poor low frequency response at 20-100 Hz. I've always held the OTs responsible for that. Today I checked the voltages at 20-200 Hz before the coupling cap, after the coupling cap and at the anode of the 1626. I was very surprised to see that the the poor FR is already present at the primary. It does not matter whether I connect a 5k or a 12k OT. Here are the voltage curves (y-axis, relative units).

Image

Schematic:

Image
Anybody any ideas?

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby Shannon Parks » Sat Nov 10, 2012 8:54 am

Three questions: What is your X and Y axis? What is your output transformer? And what is your voltage measuring device (i.e. is it True RMS)?

It should be easy to get low frequency response down to 20Hz. Linearity at 100Hz and less - due to the changing DC currents - is another issue.

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby MKJ » Sat Nov 10, 2012 9:29 am

Most of the time I use the PC for audio measurements. This time I used a Fluke True RMS Multimeter. Not a good idea to connect 300 DCV to the soundcard without any protection circuit.

The output transfromers were an Edcor XSE 5k and a vintage Telefunken 12k. The graph shows voltages before and after the coupling cap and on pin 3 (anode) of the 1626 on the y-axis and frequency on the x-axis. I was too lazy to calculate dB levels and add the appropriate frequency values.

This is the frequency response cuve of both OTs measured on the secondary. Green = Telefunken, red = Edcor:

Image

It's around 50-20000 Hz -3dB. The poor FR below 50 Hz is not a big deal. However, a flat FR would be nice. Just because. BTW, distortion increases significantly below 50 Hz and can reach 10%.

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby MKJ » Mon Nov 12, 2012 1:46 am

I've increased the value of the cathode bypass cap. Even at 2200µF I could't find any effect on the FR. This graph shows FR for 100µF and 600µF:

Image

Shannon, you posted excellent frequency response curves for (more or less) the same circuit. I don't know where these differences might come from. I can't track any drop in low frequencies before and after the coupling cap and even at the grid of the 1626 the frequency response is perfectly flat. Since you used the same OT, it's very unlikely that it has a significant effect.

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby MKJ » Mon Nov 12, 2012 4:04 am

Shannon, which Edcor type did you use? The primary inductance can vary from 3H to 10H depending on model (GXSE / XSE) and size (5-15 watts). Primary inductance could have an effect on low frequency response.

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby Shannon Parks » Mon Nov 12, 2012 5:26 am

Mike, both the Clementine 6L6 and Clementine Darling frequency sweeps were with the Edcor GXSE15-8-5K (two different sets on different amps). This past summer I designed and tested my Budgie SE - a small Clementine using EL84s - and I switched to the GXSE10-8-5K. I needed a smaller output to squeeze onto the 6X10 footprint. Anyhow, I think I get a -1dB cutoff at 15Hz and 40kHz (working from memory), though it employs negative feedback which helps.

Just to be certain it isn't a measurement or calculation issue:
1) What model is your Fluke?
2) Could you just give the raw measured voltage at 20Hz and 1kHz?

Ineloquent SE commentary: Your 10% distortion at 20Hz is normal. The real problem with these small outputs is the low inductance goes to crap during the higher current draw of your AC phase (i.e. your AC level is closest to relative 0V at the grid). So maybe 3/4ths of your cycle the output may pass 20Hz just fine (i.e. constant load and gain), but the 'top' quarter draws more current and starts to saturate your output, thus causing distortion. Your load is effectively changing all the time. So you get OK level response, but poor linearity. A huge and expensive output is the only solution. Fortunately, these little outputs still work fairly well and the low frequency distortion is a small amount of the content and is worst at max output at 20Hz.

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby MKJ » Mon Nov 12, 2012 6:38 am

Shannon, thanks for the explanation.

1) What model is your Fluke?
2) Could you just give the raw measured voltage at 20Hz and 1kHz?


It's a Fluke 85 III True RMS Multimeter. Do you mean input and output voltage?

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby Geek » Mon Nov 12, 2012 7:00 am

How odd Fluke doesn't list the frequency response when "lesser" brands do.

The true RMS response is usually poor - 40Hz to 500Hz, maybe a KHz or two, if you're lucky.

The frequency counter function isn't an indication, because the input signal is turned to square waves anyway.

Cheers!
-= Gregg =-
Fine wine comes in glass bottles, not plastic sacks. Therefore the finer electrons are also found in glass bottles.
User avatar
Geek
KT88
 
Posts: 3585
Joined: Sun Oct 21, 2007 3:01 am
Location: Chilliwack, British Columbia

Postby MKJ » Mon Nov 12, 2012 7:19 am

The true RMS response is usually poor - 40Hz to 500Hz, maybe a KHz or two, if you're lucky.

While this might be true, it doesn't matter in this setup. I measured voltages at several points in the circuit. Even if the Fluke measures rubbish this would be a systematic error. However, I checked the Fluke and the voltage readings were pretty consistent between 20 and 20000 Hz. The source was a very stable signal generator. It's very unlikely that the Fluke is the problem.

I still think that primary inductance might be a a good suspect. The XSE has a very low inductance at 3H. High quality OTs can easily have a primary inductance of 20-30H. The formula for RL filters might not be 100% appropriate in this case. However, the corner frequency for a 10k/30H OT will probably be significantly lower than for a 5K/3H OT.

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am

Postby Shannon Parks » Mon Nov 12, 2012 9:09 am

There's a real good thread on the forums where Ed Brown measured the inductance many of these Edcors:
http://www.diytube.com/phpBB2/viewtopic ... highlight=

Lots of clues in there.

Shannon
User avatar
Shannon Parks
Site Admin
 
Posts: 3764
Joined: Tue Mar 18, 2003 5:40 pm
Location: Poulsbo, Washington

Postby MKJ » Tue Nov 13, 2012 5:48 am

I'm pretty sure that the low inductance of the XSE type causes the poor frequency response. Given a plate resistance of 2500 ohms, the theoretical -3dB point would be a 130 Hz compared to 13 Hz for a 30H inductance.

Mike
MKJ
 
Posts: 20
Joined: Thu Jul 12, 2012 6:19 am


Return to clementine

Who is online

Users browsing this forum: No registered users and 11 guests