Strange Waveform - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 26th January 2015, 04:21 AM   #1
diyAudio Member
 
Join Date: Oct 2013
Default Strange Waveform

I have been building this amplifier over the last two years and once complete I will post details.
The amplifier is a digital controlled steeper attenuator based upon the design in AudioXpress by Robert Nee Dance.
After chronic power supply oscillations and many burnt chips the amplifier is based on the HA5002 buffer chip of which it uses 20! I have now got the amplifier and power supply and all the logic working but it is the attenuator that is giving the most headaches.
How can an R/2R attenuator which is nothing more than relay contacts and resistors give so much trouble you ask? Well at first I thought the resistors were the wrong type and generated some high capacitance or inductance but they are metal film Vishay components. I then decided that the veroboard was probably not a good layout but that is all I had access to so I ordered a ground-plane perforated board and used thin film SMDs that fit nicely between the pins of the relays to eliminate stray capacitance and I got exactly the same results.
The attenuator has an input resistance of 128Kohm and is controlled by an encoder and logic for a 128 step span.
When feeding the attenuator with say a 50KHz square wave bearing in mind the front end of this amplifier is an FET followed by an HA5002 buffer amp that is good for well over 1MHz bandwidth the square wave was turned into this horrible triangle wave.
So to completely make sure I'm not doing something wrong or expecting too much from a bunch of relays and resistors I setup the circuit below and got exactly the same results
This is just a function generator feeding two resistors and then connected to an an oscilloscope so as you can see I need help please.
IMG_20150125_221137 (1280x720).jpg

IMG_20150125_221229 (1280x720).jpg

Test 2.png
  Reply With Quote
Old 27th January 2015, 07:23 AM   #2
djk is offline djk
diyAudio Member
 
djk's Avatar
 
Join Date: Feb 2001
Location: USA
A 60K source impedance when loaded with any capacitance will roll off the HF as shown.
__________________
Candidates for the Darwin Award should not read this author.
  Reply With Quote
Old 27th January 2015, 07:59 AM   #3
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
128k + 10pF scope probe = 100 kHz lowpass.

Results are normal. Your square wave looks like any other square wave that went through a low order lowpass, it has no overshoot, no ringing... it is perfectly fine.

What's the problem with 100kHz lowpass for audio ?
  Reply With Quote
Old 27th January 2015, 12:48 PM   #4
diyAudio Member
 
Join Date: Oct 2013
Many thanks for the replies djk and peufeu. So I can understand this since most hifi components have nominal 50Kohm input impedance therefore you are telling me that this is normal? Does this not make ridicule of high slew rate amplifiers having great transient response?
I thought that by having a nominal input impedance of 128Kohm would be a good thing so it looks like I would be better off making the stepped attenuator to have a 10Kohm input impedance?
Some SACD players supposedly have an output frequency range upto 90KHz just because we cannot hear this frequency, being able to reproduce such high frequencies we miss a lot of harmonic beating which is why a live violin never sounds the same as a recording?
I can easily change the R/2R resistors to lower the input impedance to raise the frequency response is this my only route?
  Reply With Quote
Old 27th January 2015, 02:34 PM   #5
ilimzn is offline ilimzn  Croatia
diyAudio Member
 
Join Date: Feb 2005
Location: Zagreb
Quote:
Originally Posted by formantjim View Post
Many thanks for the replies djk and peufeu. So I can understand this since most hifi components have nominal 50Kohm input impedance therefore you are telling me that this is normal?
No - you are confusing output and input impedance.
Components have nominal (standard) input impedance of 50k but it is driven by a very low output impedance of the source. If you need to attenuate the signal, using the input impedance of the component as part of a resistive divider, then the attenuator must be in the input of the component to prevent capacitive loading by the interconnecting cable. However, the idea of using the input impedance of a component as a part of a two resistor attenuator is not a good one in the first place for several reasons, which we can discuss later in the thread. That being said, the answer to your question would be yes if this is done. Normally an attenuator will attempt to get the OUTPUT of the source to see a standard 50k resistance, but look like a lower resistance to the input, precisely to avoid bandwidth limiting.

Quote:
Does this not make ridicule of high slew rate amplifiers having great transient response?
No, as driving tham as you have shown is not the way to go in the first place. However, the fast slew rate is not really intended for the input signal, but in most cases so that the amp can react quickly enough to it's own feedback. In fact, in many cases the input bandwidth is limited on purpose due to EMI/RFI issues as you do not want your amp to become a long wave (or even short wave) transmitter should such a signal appear on the input.

Quote:
I thought that by having a nominal input impedance of 128Kohm would be a good thing so it looks like I would be better off making the stepped attenuator to have a 10Kohm input impedance?
How about a nominal 50k input impedance as is the standard?
10k is better in some respects but may be a heavyer load for the source (in particular tube circuits will be susceptible to this).
It is rather easy to make an attenuator that, loaded with a standard 50k input, will appear as a 50k load for any setting. However, it will also appear as a variable but rather low impedance as seen from the load, which will help HF response. If this is well constructed, it's really a non-issue.

Quote:
Some SACD players supposedly have an output frequency range upto 90KHz just because we cannot hear this frequency, being able to reproduce such high frequencies we miss a lot of harmonic beating which is why a live violin never sounds the same as a recording?
A violin will never sound exactly like a recording of it because a microphone is not your ear. That being said there is research into ultrasonic components of sound and it mainly has repercussions in echo perception and spatial imaging. The reason why SACD (and indeed hi res PCM) extend the frequency range far into the ultrasonic range (as defined for humans) is because of the need for low pass filtering. Agressive low pass filtering produces problems, and extending the range available makes for lower order (less steep) filters with manageable phase response.
THe standard range defined for SACD is 50kHz (-3dB) but the actual encoding is capable of usable dynamic range will into the 100s of kHz.

Quote:
I can easily change the R/2R resistors to lower the input impedance to raise the frequency response is this my only route?
I would suggest actually suggest research in to the use of L-pad attenuators.
For starters your schematic is backwards, if it is intended to be a parallel attenuator. THe input and output re reversed and the variable resistor will short the output of the generator. Also, for a 50k MINIMUM input impedance, the 60k reisistor shoud be 50k, and once the set up is properly connected, you will find that the minimum attenuation is about 3dB (which may or may not be a problem), the maximum impedance seen by the scope is around 36k and the upper cut-off is some 300kHz for a 10pF load. HOWEVER, check what the actual capacitance of your scope probe is! It may be migger than that of your amp. Also, the upper cut-off increases with increasing attenuation, which is the hallmark of such an attenuator. Honestly I am not a fan of either series or parallel attenuators as they lump various problems into one end of the attenuation range. The whole fame about passive attenuators having a 'sweet spot' re sound quality is about changing of the effective impedances depending on attenuation setting, which then influences HF and sometimes LF response. There are other attenuator schemes which do not have that problem.
Now, you might want to decrease the series resistor (which offers less minimum attenuation and better HF response) but also presents a more difficult load to the source and results in a different attenuation curve (which should ideally be close to logarithmic - and with this type, 'close' begs for a definition).
THe 50k standard was defined for a potentiometer input attenuator, and for that type, the worst case impedance as seen by the amp is at 6dB attenuation, when the pot presents a divider with equal resistors in series and in parallel with the amp input, this would be 1/2 of the pot value. THe effective series resistance when the pot is driven by a low impedance source is then the pot resistance divided by 4, i.e. 12.5k. For reasonable input capacitances the HF cut-off is well into the 100s of kHz range and normally not an issue. THis standard was not just pulled out of a hat.

Last edited by ilimzn; 27th January 2015 at 02:59 PM.
  Reply With Quote
Old 27th January 2015, 05:41 PM   #6
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Quote:
Originally Posted by formantjim View Post
Does this not make ridicule of high slew rate amplifiers having great transient response?
Slew rate limiting is due to a limit on the current that is available to charge the compensation capacitor inside the amplifier. It is clipping of the input stage (on most topologies).

Since distortion of the input stage is not corrected by feedback, it is preferable to operate the input stage far away from clipping, where it is more linear.

So, a higher slew rate amplifier which is actually using only a small part of its slew rate for audio will distort less because it puts less strain on the input stage (all other things being equal, which is never the case ) ; but you could do the same with a low (but sufficient) slew rate amplifier having a very linear input stage, etc.

You don't need ridiculously high slew rate or high bandwidth amplifiers for audio. Simply, it is a characteristic of the amplifier that is correlated to its audio qualities, but not directly.

So, it is not stupid to put a lowpass on the input of a very fast amplifier, since this lowpass filter will :

- Not affect the audio signal (if the cutoff is high enough),

- Remove the HF noise picked up by cables (cellphones...)

- Remove the HF noise emitted by the source (for example I have a E-MU soundcard here, it emits lots of HF crap in the output, switching DC-DC residue, a healthy bit of DAC clock leaks through, really it is full of garbage). On such a source, adding a lowpass will reduce intermodulation distortion in the amplifier due to the HF crap, which tends to make it sound different.

Quote:
Originally Posted by formantjim View Post
I can easily change the R/2R resistors to lower the input impedance to raise the frequency response is this my only route?
It is more about the input impedance of the amplifier you put afterwards. If it has a lowpass filter at the input, or the parasitic capacitance of some opamp input stage, you need to know that, because the output impedance of your attenuator will matter here.
  Reply With Quote
Old 28th January 2015, 01:34 AM   #7
diyAudio Member
 
Join Date: Oct 2013
Thank you both peufeu and ilimzn for your extremely detailed replies!

This is exactly why I signed up to this great forum the wealth of knowledge and explanations are exceptional!

I think I may have misled you with my circuit diagram earlier the attached diagram is what this attenuator feeds courtesy of Robert Nee Dance article in Audio express "Willow FET". As you can see it is an FET front end with extremely high input impedance and a buffer chip.

Since the excellent answers caused me to research (I took the hint!) a bit more I have since discovered that his original design for the stepped attenuator was 12Kohms and I had simply overlooked this and gone with my heart so to speak with 128Kohms with the disastrous results as we have seen.

I'm going to do further tests with different resistor values and see which gives the best results and will report my findings.

"How about a nominal 50k input impedance as is the standard?
10k is better in some respects but may be a heavier load for the source (in particular tube circuits will be susceptible to this).
It is rather easy to make an attenuator that, loaded with a standard 50k input, will appear as a 50k load for any setting. However, it will also appear as a variable but rather low impedance as seen from the load, which will help HF response. If this is well constructed, it's really a non-issue."

This is an excellent suggestion and since this R/2R ladder uses mutiples of 2 and I have constructed using 16kOhm resistors then either 64K or 32K maybe my best bet.

Again many thanks I learn something every time I'm on this forum!Preramp input.png
  Reply With Quote
Old 28th January 2015, 09:19 AM   #8
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Input impedance of this circuit may not be quite as high as you think, as Miller effect means the effect of Crss and associated nonlinearity would increase by circuit gain. Cascoding the FET might be worth it. Seems worth looking at in simulation.
  Reply With Quote
Old 28th January 2015, 11:50 AM   #9
diyAudio Member
 
Join Date: Oct 2013
Thanks sgrossklass the circuit works just fine and it has been my struggle with the change in the square wave that brought me to this question.
Doing DIY and measuring things and getting unexpected results has forced me to research this subject more deeply and has now given me an appreciation of how various elements in the audio chain can have an effect on the sound we hear.
This simple thread Impedance question [Archive] - AudioKarma.org Home Audio Stereo Discussion Forums I came across explains things very clearly.
ilimzn commented on the output of tubes and the above link also has helped me understand the loading effects input impedance works.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Triangular waveform dead rat Class D 27 11th November 2008 06:54 PM
Waveform analysis Conrad Hoffman Equipment & Tools 1 23rd March 2008 03:53 AM
Funny waveform! ifrythings Power Supplies 19 1st April 2006 12:49 AM
Brazil, and some strange laws, not too much strange! destroyer X Everything Else 0 11th September 2004 01:38 PM


New To Site? Need Help?

All times are GMT. The time now is 09:49 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright 1999-2015 diyAudio

Content Relevant URLs by vBSEO 3.3.2