What makes the most sound improvement?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yet another newbie lookibng for assistance...!

Ok, so I recently threw together my first chip amp (LM3876) , and the bundle of wire and solder actually produces sound!! I was thrilled !! DC offset was approx. 5mV

I started with a minimal circuit, with no capacitors apart from the 1000uF supply caps. ALL the components used are cheapo quality, with the aim of making some audible improvements soon.

Experimentation so far:

Audible hum, and occasional radio signals heard when no signal present. --- I added a 220pF disc ceramic (as per National datasheet) across the input pins of the chip, and made a massive improvement. I can still hear a very small amount of hum with my ear next to the speaker.

Capacitively coupling the input. --- I added a 22uF bipolar electrolytic in the signal path. -- From wiper of the pot. to the input of the amp. This seems to create lots of nasty hum, and the chip got really hot (it's usually quite cool). I presume this is due to oscillations. But I'm not sure why they occured. When I removed the cap., everything went back to normal. Can anyone explain this?

So where do I go from here??? The sounds is nice at the momment, but possibly a little harsh at the top end. I'd like to try swapping some components to make improvments in the sound, but am not too sure where to start. I'm planning to get some Panasonic FCs for the supply caps, but where to from there??? Bigger transformer? Bigger bridges? Exotic resistors?? Maybe add some kind of high frequency compensation network in the feedback loop?

I'd like to hear some of your views as to what makes the biggest difference to the sound.

thanks,

Steve

Current test setup is:

POWER SUPPLY (was built before I started this, so I used it)
160VA Dual 16V toroid.
2 x 6A bridges
10,000uF on each rail -- gives +/- 25V with no load

REST oF SYSTEM
Portable CD player source
20k Carbon track pot
22k carbon f/b res.
22k from mute pin to V-
1k carbon res. to ground.
LM3786 chip
2 x 1000uF 25V cheapo electrolytics
All ground connection return to a single point.
 
Capacitively coupling the input. --- I added a 22uF bipolar electrolytic in the signal path. -- From wiper of the pot. to the input of the amp. This seems to create lots of nasty hum, and the chip got really hot (it's usually quite cool). I presume this is due to oscillations

Seems like the cap breaks the dc ground reference on the positive input - all sorts of bad things may happen. Why don't you move the cap ahead of the pot? Why use a cap anyway?
If the sound is harsh on top it may well be due to the source. A buffer may well mellow it down.
 
analog_sa said:


Seems like the cap breaks the dc ground reference on the positive input - all sorts of bad things may happen. Why don't you move the cap ahead of the pot? Why use a cap anyway?
If the sound is harsh on top it may well be due to the source. A buffer may well mellow it down.


I wanted to try the input cap just to see what any differences may be. It was easier to connect it to the wiper, as my RCA socket for the input signal is soldered directly to the 'top / input' leg of the pot.. so I was just being lazy really. I think you might be right about blocking the dc ground.I'll try inserting the cap before the pot, and will check the results again.
 
re: Buffer!

Nuuk said:
BUFFER

See

HERE . ;)

Nuuk..

Hmmm, Have looked at your webpage and will try building a buffer. - If I can get similar results to you, it looks like I may achieve the kind of sound I'm looking for. Those OPA627 Op Amps are a little pricey for me (NZ$50 ea. from RS, when my LM3876 only cost NZ$20!) so I might experiment with a cheaper alternative first. Is the NE5534AN any good? Does anyone have any other suggestions?

Steve
 
Is the NE5534AN any good?

With all the other more recent opamps around, many people poopoo the good old NE5534 but the truth is, it is still a very good opamp, especially when biased into Class A.

Last Tuesday I had the chance to listen to a pair of these used on the output stage of a modified CD723 and they sound extremely good.

So yes, start with the 5534 BUT as I remember they cannot be used like the OPA627 in my circuit, ie they are not stable at unity gain so check out the data sheets for the correct implementation.

If you use sockets for the IC's, you can easily try other ones later.
 
NE5534 will be the test IC.

Nuuk said:

it is still a very good opamp, especially when biased into Class A.

Can you give me a rough idea of how this is achieved?


ie they are not stable at unity gain so check out the data sheets for the correct implementation.

If you use sockets for the IC's, you can easily try other ones later.

From the datasheet, it looks like you're right - the NE5534 is only stable with gains greater than 3.

"The op amps are internally compensated for gain equal to, or higher than, three. The frequency response can be optimised with an external compensation capacitor for various application (unity gain amplifier, capacitive load, slew rate, low overshoot, etc.)"

Although the datsheet doesn't provide application notes, a quick Google for 'NE5534 unity gain' turned up AN142.pdf from Phillips Semiconductor. PLenty of usuable information and application circuits there!

It looks like unity gain is achieved with a 22pF cap. and a resistor in the feedback loop.

So I went shopping yesterday lunchtime, and came back with a pair of NE5534s, sockets, caps, and resisitors. I've also got strip (Vero) board, but was wondering if it would be OK to prototype with breadboard???? I know some circuit have issu wies on breadboard due to the inherent capacitance of the contact strips, but I'm not sure if that is just for high speed digital stuff.

Thanks for the help so far.....now if only I could stop thinking about amplifiers, and get on with some work........

Steve
 
Hi Steve, yes it's addictive alright! :hot:

There are several methods of biasing into class A (try a search) but probably the simplest is to use a 7K5 resistor from the output to the negative rail. This is assuming that you use +/-15 volt rails.

Just for the hell of it, I may try the 5534 like this (class A with gain) and compare it to the OPA627's I use at present.
 
Nuuk said:
Hi Steve, yes it's addictive alright! :hot:

There are several methods of biasing into class A (try a search) but probably the simplest is to use a 7K5 resistor from the output to the negative rail. This is assuming that you use +/-15 volt rails.

Just for the hell of it, I may try the 5534 like this (class A with gain) and compare it to the OPA627's I use at present.

Yep, I'll be aiming to use +/- 15V. I'll try the resistor approach. I also found this:

http://tangentsoft.net/audio/opamp-bias.html website that is also very informative. I may also try use the CRD (current regulating diode) method as it appears to be a good compromise between simplicity and variance in the bias current.

Of course I'll be interested to hear your about your listening test with a NE5534 replacing the OPA627. - Especially since the OPA's are 10 times the price !! :bigeyes:

I'll be knocking something up on the breadboard tonight.

Steve
 
Suggestions

……but where to from there???

Steve,

Some suggestions based on my own experience.

General…
Replace the volume pot with an attenuator (shunt type).
Upgrade the electrolytics.

With reference to reducing harshness.
Cheap electroylics sound congested and harsh, use film types where possible, preferably Jensen paper-in-oil.

Buffer…
I have repeatedly tried the OPA627 (and many other OPA/AD op-amps) in two different circuits (in, and not in, class A). I have consistently formed a preference for the AD797, NE5534, NJM2114 and the LM6171 due to their more natural presentation of detail. For striking dynamics, excitement and thrills the OPA627 is perhaps a better choice, along with many other FET input op-amps, though to my sensitive ears most of the FET types sound hard and edgy to some degree. The AD6810 and the AD825 are exceptions. For tonal accuracy, smoothness and fine detail try the LM6171 in class A, though fit a 500nF film cap across its supply pins, though don’t expect much in the way of ambient warmth. For that try the NE5534 which improves on all of the above with respect to fine ambient detail, or the AD797. For many I am sure the NE fails to provide excitement and it certainly sounds less dynamic than most, but it somehow manages to extract all that is good in a recording whilst removing the rest. Or should that read ‘it doesn’t add anything of its own’. For best results perhaps you could omit the NE5534 compensation capacitor and configure the circuit for some gain. The NE5532, which has the compensation capacitor built-in, sounds less good in my circuit.

Tim.
 
Re: Suggestions

TimA said:
……but where to from there???


Replace the volume pot with an attenuator (shunt type).


You mean a stepped attenuator...or something else?


Upgrade the electrolytics.

I'll be ordering some Panasonic FC's for the supply caps soon.

With reference to reducing harshness.
Cheap electroylics sound congested and harsh, use film types where possible, preferably Jensen paper-in-oil.


Buffer…
For best results perhaps you could omit the NE5534 compensation capacitor and configure the circuit for some gain. The NE5532, which has the compensation capacitor built-in, sounds less good in my circuit.

Tim.

I quickly built a buffer onto breadboard last night, using an NE5534 and decided to follow the circuit Nuuk has on his website, without any unity gain compensation. It worked (ie I got sound) , but I haven't had a chance to listen to it properly yet. It'll have to wait 'till I get back from my hols to play with it more.

Steve
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.