• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Tube Amplifier Stability Test

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I’m building a single ended 5W amp with an EL34 in Ultra-linear mode, using about 6dB global negative feedback (GNF).

The output transformers are Chinese and have very good low frequency performance, but an obvious resonance at about 70kHz. Although my preference is for Edcor, their delivery times and freight to Australia is a real killer.

To prevent oscillation in the output stage I’ve found it necessary to add a 1nF cap between UL screen and B+ transformer connections. That seems to have solved that issue.

I’m now looking at an easy way to test the stability margin of the amp with 6dB GNF. The unity gain-bandwith is about 150kHz, so I’m concerned about the 70kHz transformer resonance.

I’m connecting a resistive load and increasing the amount of GNF until the amp oscillates. Oscillation occurs if I increase the feedback by 11dB. Is this a valid test and, if so, is 11dB enough? Should I be using a complex impedance load such as 6 Ohm in series with some R+C ?

Thanks...:)
 
If you slow down your amplifier and limit the frequency to 20kHZ, you will save on wasted power with frequencies we can't hear. The cap shunting the output transformer is an excellent way to snub the top end. Williamson used a 470p in series with a 1k2 in parallel with the anode load of the first triode. That kept the driver at bay.

Low frequency is what the human ear finds pleasing and lots of iron in the transformer helps with that. Don't over do the NFB, you may make it worse.

I would have thought there were transformer manufacturers in Australia , like we have in UK that can do a better job than Chinese manufacturers.

The pictures are of my Type 807 power amplifier. All transformers were made in England to my design and locally at a cheaper price than anywhere else I could find. I made 3 of these, two sold straight away and the first one I kept for my home use.
 

Attachments

  • IMG_20160102_181306 copy.jpg
    IMG_20160102_181306 copy.jpg
    503.4 KB · Views: 446
  • IMG_20160106_110735 9copy.jpg
    IMG_20160106_110735 9copy.jpg
    640.7 KB · Views: 450
  • Top view copy 2.jpg
    Top view copy 2.jpg
    495.3 KB · Views: 442
I'm learning how to use this site.. so sorry if I muck it up.

"slow down your amplifier and limit the frequency to 20kHZ you will save on wasted power with frequencies we can't hear...."

Its always puzzled me why people value an HF response up to 50 kHz or more. I always put it down to transient performance must be important, but I have never done any listening tests comparing a 20kHz response to say 50kHz.

I would have thought there were transformer manufacturers in Australia , like we have in UK that can do a better job than Chinese manufacturers.

I'm thinking of making this stereo amp into a kitset with a laser cut chassis, so I've been looking for an output transformer for less than A$50 landed cost, which I can do with Chinese transformers. There are a couple of people making transformers in Australia but they are high-end products.

Your pics look good. Attached is one of mine. Its a circlotron with 4xKT120's per channel and (few hundred Ohm output impedance of the circlotron cct means I can use thin hook-up cable for the leads from the amp and mount Lundhal output transformers at the speakers).
 
A schematic of your amp saves 1,000 words. Sometimes one of the others will catch something about the amp that would otherwise go unnoticed. It could be a fix, a simple improvement, an unnecessary part, or a mistake in the schematic, etc.

An RC snubber from the driver plate to ground might be in order, schade feedback might be good, etc. Wrapping global negative feedback around 2 stages and an output transformer may be difficult to adjust. It can be a problem when you stabilize for a non inductive resistor, but wonder what happened when you connect a loudspeaker with impedance and phase all over the map versus frequency.
 
Last edited:
Here is the circuit - I have simplified it by removing the input switching ect. It uses a constant-current-draw-amplifier (CCDA) per John Broskie's TubeCad, followed by a conventional SE output stage. I like the CCDA because it drives the output tube with a fairly low impedance and puts the second triode to good use. The plate voltage is intentionally low.
 

Attachments

  • 4.5W SE AMP - PHIL WAIT.pdf
    16.6 KB · Views: 182
Disabled Account
Joined 2009
Here is the circuit - I have simplified it by removing the input switching ect. It uses a constant-current-draw-amplifier (CCDA) per John Broskie's TubeCad, followed by a conventional SE output stage. I like the CCDA because it drives the output tube with a fairly low impedance and puts the second triode to good use. The plate voltage is intentionally low.

A small capacitor of 100 ~ 150pF at the grid of the power tube may help too. Langford-Smith (of Raditron fame) recommends 1 ~ 2.2nF and 47 ~ 220 in series between plate and g2 when in UL.

Quality of OPT has everything do with this and it depends on the OPT. In a Dutch book on designing amplifiers from way back there is a schematic where they used 5nF in series with 20K across the primary of a non-UL OPT for an EL84.

I do not like to put too much credence on a square wave, it can hide some other stuff: a "perfect" square wave and you might find that sometimes the DC resistance is pretty high. I like to follow the square wave testing with a triangle and check for irregularity / rounding at the peaks.
 
Disabled Account
Joined 2009
No need for elaborate calculations or equipment. I use a simple USB signal signerator / oscilloscope (Picoscope 2204A) and a variable capacitor and a potentiometer. After getting the "perfect" square wave just measure the values.

Courtesy of Mr Smith.
 

Attachments

  • feedback adjustment.zip
    220.8 KB · Views: 55
trobbins,

1. Some search for a subject or part.
They find a thread, and do not pay attention to how old the thread's last post is.

2. Some search for a category of current up-to-date posts, like the recent Tubes / Valves
threads.
That works (Unless they just did a search of a recent post, but where an old thread was re-activated by those who use method # 1 above,
and then added a new posting to it, which re-activates that very old thread they "just found".
 
Last edited:
Yeh I've posted to a an old thread before without realising the dates.

The recent posts sort of infer some advise back to the OP to use a different method - which after 2 years may well fall on deaf ears - hence my comment in case AmadeusMozart was anticipating a response from the OP.

As a general comment, I note that the OP was concerned about a particular test he was making and how credible that test method was to give him confidence about his stability margin. The new posts don't really advance that issue, and rather just dismiss the OP's effort and test as too elaborate.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.