Old amps need reining in

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'd like to work out the correct impedance for a new volume pot in my system. Actually, I hope someone else can work it out because I'm confused after reading far too much here and not listening enough in physics.

I am running two Quad 303 amps straight off a PC & DAC, so no preamp, just a passive volume pot I hid in an FM3 box as a temporary solution a decade or two back. It works OK but doesn't take the volume down low enough, meaning I have to use the Windows audio stack to keep it in check. I'd prefer to bypass this and send the full signal out, unmolested by Microsoft engineers.

The numbers I know are:

2 x Quad 303 amp, input impedance of 22K ohms each (one for each L&R channel). Not sure if using them like this effectively halves that value?

DAC output impedance 50 ohms

I'm looking at a choice of a 10k, 50k or 250k log potentiometer.

Advice gladly taken.

Thanks.
 

Attachments

  • IMG_20191216_095042.jpg
    IMG_20191216_095042.jpg
    449.3 KB · Views: 280
Last edited:
Administrator
Joined 2007
Paid Member
Welcome to diyAudio :)

I would go for a 10k in this situation as any (modern) solid state source component such as a DAC will easily drive such a load.

Where have you got your 22 ohms input impedance from? It is way higher than that on a 303 :)

(edit... I see post #1 has been edited to reflect the error)

The Quad 303 is pretty sensitive (doesn't need much signal) and I suspect your issue of poor control is actually one of simply supplying to much signal... so use a resistor in series with the feed to each pot to cut the level. Try something like a 10k if using a 10k pot and see if that helps. You can go higher or lower.

Pots can be poor in tracking between left and right channels at low volume and so I would suggest using a linear pot, not a log and faking a log law by addition of a resistor from the wiper to the ground pin.

Have a read near the bottom of the page here:
Potentiometers (Beginners' Guide to Pots)
 
AX tech editor
Joined 2002
Paid Member
I'd like to work out the correct impedance for a new volume pot in my system. Actually, I hope someone else can work it out because I'm confused after reading far too much here and not listening enough in physics.

I am running two Quad 303 amps straight off a PC & DAC, so no preamp, just a passive volume pot I hid in an FM3 box as a temporary solution a decade or two back. It works OK but doesn't take the volume down low enough, meaning I have to use the Windows audio stack to keep it in check. I'd prefer to bypass this and send the full signal out, unmolested by Microsoft engineers.

The numbers I know are:

2 x Quad 303 amp, input impedance of 22K ohms each (one for each L&R channel). Not sure if using them like this effectively halves that value?

DAC output impedance 50 ohms

I'm looking at a choice of a 10k, 50k or 250k log potentiometer.

Advice gladly taken.

Thanks.

Can you show a circuit diagram? Are you sure the bottom of the pot is actually at ground? It is very unusual that you can't turn it down enough. It will never be absolutely zero, there's always some residual resistance, but with any reasonable pot -60dB should be achievable. And that's very, very soft.

Jan
 
OK, the fog is starting to clear. I think you're on to something with the ground as I pulled everything around a while back, in the most litereral sense, and had a few problems afterwards with uneven balance on the volume. This isn't something I'm proud of but I may have fixed it by giving everything a good shake.

I won't test your patience by trying to draw a circuit diagram but suffice to say the stereo volume pot has been spliced into the two cables coming out of the DAC, carrying the L & R signal. Each side then goes on to a seperate 303, the driver boards powering either the high or low range of one KEF Concerto speaker. One day all hifis will be built like this.

I may have thought about the earth when I put the pot in all those years back but certainly haven't since. I'll take a look.

Many thanks.
 
I won't test your patience by trying to draw a circuit diagram but suffice to say the stereo volume pot has been spliced into the two cables coming out of the DAC, carrying the L & R signal.
Sounds like you have simply wired the pot in series, which is not how it's normally done. Typical volume pot wiring between signal and ground should be good for at least 40-60 dB of attenuation, I can't imagine you need more than that. AC gain for the 303 should be roughly -40, or 32 dB.

The 10k pot you ordered seems like a good choice.
meaning I have to use the Windows audio stack to keep it in check. I'd prefer to bypass this and send the full signal out, unmolested by Microsoft engineers.
How much more do you need? Even if there is no PGA involved, I would be entirely unconcerned about something like -20 dB of attenuation.

What I would be much more concerned about is the inevitable ground loop, with both the PC audio output ground and the amplifier ground being connected to PE (both being IEC Class I devices). Results could be less than pretty under load. A ground loop isolator thingy between PC and volume control seems rather advisable indeed.
 
Pots can be poor in tracking between left and right channels at low volume and so I would suggest using a linear pot, not a log and faking a log law by addition of a resistor from the wiper to the ground pin.

On the contrary, I find the fear of mistracking unjustified and actual tracking on modern pots quite accurate.
Joys of automated manufacturing and ISO 9000 something quality controls I guess.

Why wouldn´t they track well? after all they are made out of stacked *linear* sections :D

I find curve faking very poor compared to the real thing, even the commercial version, and that the worst departure from that is precisely at lower settings where faking provides way higher levels than desired .... the opposite of what OP needs.

I´m copypasting Rod Elliot´s curves which I would not normally do, but sadly is needed here so this error can be clearly seen side by side (actually top-bottom he he) because just referring to an outside page does not seem to be clear enough, since; they are quite separate, there´s lots of text in between ,**other pot circuits** and to boot they do not use the same scale (% vs dB :confused:):

Real and Commercial Log:
pots-f4.gif


Fake log:
pots-f8.gif


the error (clearly shown on graphs) is way higher than hinted at in text where it´s dismissed as irrelevant.

Notice that with pot on "5" or halfway rotation.
, Commercial Log provides 20dB attenuation (10% input voltage) which is expected (duh!!! , as if it were designed to approximate the real Log curve :rolleyes:) while Fake Log provides 14dB attenuation, a GROSS 6dB error.

An error which INCREASES at lower pot settings, where the upwards bent curve is even furher away from the downwards bending required in true Log.

In fact (just LOOK at the curves), so called Fake Log actually behaves as ANTI-Log at lower settings.

Even "2 linear sections cascaded" as seen on Commercial Log matches it better, at least does not further deviate the wrong way.

Hope seeing both curves side by side helps detecting the shortcomings better.

Again kudos to Rod Elliott and his ESP site for the incredible ammount of real, practical information he carries.

Although in this small area I beg to differ.

PS: as extra data, I have actually manufactured pots at an ex-Philips equipped Factory, go figure :eek:
And process is simple: a long strip of Pertinax is coated (with stainless steel rollers) with a special "graphite dust carrying phenolic varnish" which is then dried and cured in a temperature controlled oven to provide a resistive layer.

Linear pots have a full width single coat, Log ones have two, side by side and very slightly overlapping at the center line.

You can mix any base paint combination you like, but of course Log ones have , say, 10k 0>5 and 90k 5>10 so in this case 100k Log (which I used most) with 10% (-20dB) centerpoint.

After coated strip is cooked, dies punch out of it the required Omega shape needed for pots.

If you want extra-matching, you use tracks cut from the same strip, which match by definition (similar to same-die semiconductors)

Making Antilog ones is *easy*, you just punch them "upside down" .

Given I had personal access to machinery, I also made S curve pots (great for equalizers and NFB type gain controls) which is basically a higher resistance center layer and 2 lower resistance edges (Log/Antilog in a single track).

I HATE all this Globalized "lets move all Manufacturing to China" approach, which although offering somewhat cheaper product makes you lose direct contact with Manufacturer ... unless you order 1000000 parts of course :mad:
 
In fact (just LOOK at the curves), so called Fake Log actually behaves as ANTI-Log at lower settings.
No it doesn't. It returns to linear at low volumes but never any more than that.
attachment.php


Mind you, it is clear that "fake log" is of little use if you need more than 25-30 dB of control range, which in turn tends to be covered plenty well by typical log pots already, so yeah. Looks to be most promising for a limited range (like maybe 20 dB) level trim where the resistor in series keeps the lower end of the curve in check.

BTW, I'd recommend not using an accent (´) for an apostrophe. It's garish and makes you look like a fool.
 

Attachments

  • Pot-FakeLog-1.png
    Pot-FakeLog-1.png
    32.9 KB · Views: 196
Last edited:
Administrator
Joined 2007
Paid Member
On the contrary, I find the fear of mistracking unjustified and actual tracking on modern pots quite accurate.
Joys of automated manufacturing and ISO 9000 something quality controls I guess.

Why wouldn´t they track well? after all they are made out of stacked *linear* sections :D

Perhaps it is lingering bad memories of cheaper pots from years ago that is influencing me here although I do think a good linear pot takes some beating. I have personally experienced bad tracking and it is very annoying as the central image shifts one way or the other as you listen at low level.

Having a good gain structure helps as well, something I never appreciated at the time and so I would urge Steve to consider adding some series resistance to the feed into the pot if he finds 'all the action' is happening in the first quarter turn or so of the control.

I'm pretty sure that's what I did. I've thrown some money at a new one to atone for it, and having learnt a lot of new things investigating this, I've got excited and have gone for a "stepped ladder attenuator". Just need to fit it properly this time.

:) If it was in series then even a 100k pot would only cut the gain down to around 16db or so... in other words still really loud.
 
Member
Joined 2010
Paid Member
Speaking of Quad 303s and volume controls......, the article linked below looks to be just the ticket and Keith Snook goes into some interesting details specific to the Quad in the course of his upgrade. The added ALPS blue pot, or even the popular copies of them, are much smoother and better tracking than almost any other volume pot. that isn't priced like gold dust. ~ QUAD 303 Amplifier modified for general purpose Lab amp ~
 
:) If it was in series then even a 100k pot would only cut the gain down to around 16db or so... in other words still really loud.

Yes, that describes it. I'd thought the way it was physically fitted, concealed in a scrap FM3 with the original tuning knob doing the turning, meant that I wasn't able to use the adjustment full range. It looks like I made a more fundamental error than that. I was thinking I'd got the pot from Wilmslow Audio but now I have a vague recollection of someone loaning me a huge RS catalogue and a pin.
 
Administrator
Joined 2007
Paid Member
An easy mistake to make :)

One of the outer tags of the pot goes to ground, the signal feeds into the other outer pin and the signal is taken from the middle (the wiper). Get the two outer ones back to front and the pot works backwards of course.

So when you wire it, turn the pot to minimum volume (fully anticlockwise) and then see which outer pin reads as shorted to the middle one. That outer pin is then your ground one.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.