CA 340A SE LM3886 based amp - Upgrade advice please.

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

I own a Cambridge audio 340A SE Integrated amplifier. I have allready modded the Preamp stage and apart from a minor buzz problem (this will be fixed soon) the sound is getting really good. I want to turn my attention to the power stages but I'm afraid this is where my knowledge and experience runs out.

As I understand it the poweramp stage is basically a Gainclone, it uses the LM3886 Chip. I have read good things about the DIY Gainclone and thought it would be a good idea to see what mods could be made to improve mine. It looks fairly simple with just a few resistors, power rail decoupling capacitors, and a feedback capacitor. So far I have replaced the input caps with Mundorf MKP's. I would like to know what are the 'tried and tested' caps to use for the decoupling and feedback, and if there are any gains to be had with replacing resistors. Also some power supply modifaications are planned. Currently the amp has +/- 10'000uF 50v smoothing caps. Im thinking of replacing these with Mundorf M-Lytics of the same value (possibly slightly higher capacitance?). The Diodes are 1N5402 which I plan to upgrade also. Are the Hexfreds a good choice, if so which?

This is the schematic: (C23 is not fitted as standard)


l_152bf73fe93f45ceb76dd57829470be0.jpg


Any ideas?
 
C3 is far too big,
C22 is unmatched to C3 and is too small.

Set your input MKPs and the 22k to give 80 to 100mS RC time constant. Any Polypropylene here is better than any electrolytic. Try the cheapest PP you can find before spending big money on a Mundorf. 3u6F to 4u7F would do.

Set C22 and the 1k0 to give 140 to 200mS RC time constant. 150uF would do.

Does your preamp have a DC blocking cap at it's output? Probably not since it is integrated with the power amp. If it did you have to take account of both caps to calculate the RC time constant of the high pass power amp input filter.

Add a RF attenuating filter to the power amp input.
Check there is a RF attenuating filter at the RCA inputs or at the preamp input.
 
Many thanks for that . Im afraid however that I dont understand this time constant stuff? Im well capable of replacing components and understand most of your post though. The preamp (LM4562 +6dB gain stage) has DC bolcking caps on its input, upgraded to Mundorf 2.2uF MKP's. The output feeds into the tone/Volume circuit via another 2.2uf MKP. Then the feed from the pot goes on to the powestage (LM3886) via another pair of Mundorf MKP's. Someone advised dropping R53 from 270k to 22k for an improvement in PSSR. Again, I dont know what PSSR is/means?

To be honest Im hoping to get some simple advice, I know its kind of lazy but I havent got time to study amplifier basics. :confused:;)
 
While I appreciate your knowledge, I find it unhelpful that you cannot explain you earlier post in laymans terms. Im sure its a very good post and I would be quite certain of the improvements it would bering to my amp if only I knew what it meant. I assume you are talking about oscilloscope readings and the like. I do have a scope but I havent got around to learning how to use it. To be honest Im only looking to upgrade caps and resistors. Its interesting however that you suggest tweaking the values....

If I can ask one more question to you sir: Would you suggestions above bring a definate improvent in sound quality?
 
The RC time constant is really exactly what it sounds like. The input cap and resistor to ground make a high pass filter. The time constant here is the resistance multiplied by the capacitance - RC. The corner frequency of the filter is 1/(2*pi*R*C), and you want this to be far away from any audible frequencies.

A 10uF polypropylene cap is prohibitively large and expensive, but fortunately, a smaller 3.6uF will do fine (I would even use 2u2 if it's all i had). For instance 1/(2*pi*22000*0.0000022) = 3.3Hz which is fine, 1/(2*pi*22000*0.0000036) = 2Hz which is better, but neither should cause any problems
 
valleyman said:
The RC time constant is really exactly what it sounds like. The input cap and resistor to ground make a high pass filter. The time constant here is the resistance multiplied by the capacitance - RC. The corner frequency of the filter is 1/(2*pi*R*C), and you want this to be far away from any audible frequencies.

A 10uF polypropylene cap is prohibitively large and expensive, but fortunately, a smaller 3.6uF will do fine (I would even use 2u2 if it's all i had). For instance 1/(2*pi*22000*0.0000022) = 3.3Hz which is fine, 1/(2*pi*22000*0.0000036) = 2Hz which is better, but neither should cause any problems


Thats good then. As the DC offset is so low on the LM4562 output I'll be removing a few of the siganl decoupling caps. Possibly I can get away with just a single 2.2 MKP on the Poweramp input.

Im very interested on the PSU above. It seems to be ideal in terms of size and It's dual mono configuration combined with the high quality parts should be good. I need to confirm its suitable first.
 
mikesnowdon said:
I could probably fit one of these in he chassis:

http://www.chipamp.com/supply.shtml

An externally hosted image should be here but it was not working when we last tested it.


Could this PSU be run of a single 23v transformer winding and still give +/- 28v into each output, or would it divide that to 2 x +/- 14v?

The standard Cambridge Audio toroidial has 23 - 0 - 23 windings for he power amp supply.
 
Lacks a lot.

I am surprised how close this is to the National data sheet circuit. All the time constant info is correct, a 2.2 to 3.3 ufd input cap and change the C22 to 150 - 220 ufd.
It would really benefit from a more robust power supply. Something like 10K - 20k ufd per rail. Then maybe smaller value caps on the board, 100 - 330 ufd for less inductance.
But in my experiance, a small low impedance (Pana FM is perfect) electro soldered right on the bottom of the board. 10 -33 ufd works great. Solder from 1-5 to 7 and 4 to 7.
I do not know if it is the improved decoupling, or the faster current delivery of a small cap with very short leads. But the LM3886 ALWAYS sound better to me afterwards.
If people like these stock, the potential is there to get much more.


George
 
mikesnowdon said:
Could this PSU be run of a single 23v transformer winding and still give +/- 28v into each output, or would it divide that to 2 x +/- 14v?

The standard Cambridge Audio toroidal has 23 - 0 - 23 windings for he power amp supply.
no, you need 0-23Vac, 0-23Vac to run the dual bridge rectifiers in that pictured PSU.
http://www.chipamp.com/images/ps.gif
However you can remove 4 of the diodes (and insert jumper leads) to give a single 4diode bridge rectifier and then you can use the 23-0-23Vac transformer to feed that PSU.

PS,
if you do your share of the homework, then I'll play ball.
Admit you're lazy and do nothing for your self and you won't get any respect from me.
 
AndrewT said:

no, you need 0-23Vac, 0-23Vac to run the dual bridge rectifiers in that pictured PSU.
http://www.chipamp.com/images/ps.gif
However you can remove 4 of the diodes (and insert jumper leads) to give a single 4diode bridge rectifier and then you can use the 23-0-23Vac transformer to feed that PSU.

PS,
if you do your share of the homework, then I'll play ball.
Admit you're lazy and do nothing for your self and you won't get any respect from me.


Thanks for replying. You have confirmed what I suspected........
 
Re: Lacks a lot.

Panelhead said:
I am surprised how close this is to the National data sheet circuit. All the time constant info is correct, a 2.2 to 3.3 ufd input cap and change the C22 to 150 - 220 ufd.
It would really benefit from a more robust power supply. Something like 10K - 20k ufd per rail. Then maybe smaller value caps on the board, 100 - 330 ufd for less inductance.
But in my experiance, a small low impedance (Pana FM is perfect) electro soldered right on the bottom of the board. 10 -33 ufd works great. Solder from 1-5 to 7 and 4 to 7.
I do not know if it is the improved decoupling, or the faster current delivery of a small cap with very short leads. But the LM3886 ALWAYS sound better to me afterwards.
If people like these stock, the potential is there to get much more.


George

Thanks George.

A freind who dosent mind helping out had a look at that dual bridge supply for me, wont work without a new transformer as I suspected. We both decided the best option would be to upgrade the standard 10'000 uf caps to 15'000uf (Probably Mundorf M-Lytic), with a smaller polyprop bypass under the board and install Hexfred diodes. Simple and easy to do so I'm happy.

Looking at the schematic I see there are allready decoupling caps. Are you suggesting to replace these with Pana FM's soldered direct to the chips pins? Sounds good allready. And thanks for confirming the input and feedback suggestions. For the feedback cap is there a favorite type/brand of cap to use?
 
Re: Re: Lacks a lot.

mikesnowdon said:
upgrade the standard 10'000 uf caps to 15'000uf (Probably Mundorf M-Lytic), with a smaller polyprop bypass under the board and install Hexfred diodes. Simple and easy to do so I'm happy.

...For the feedback cap is there a favorite type/brand of cap to use?
The important thing is to get the right value and the apropriate component type. Throwing money at brand names means wasting it.

Check, if the rectifier diodes are snubbered with small capacitors, varistors or RC circuits. If that is the case, forget about the Hexfreds. Else consider to snubber the rectifier and forget the Hexfreds as well.
Then again the 1N5402 is already a fast-switching diode, so you may even get away without snubbers.

If you often listen loudly, you may consider to replace the diodes with others of higher current rating. That's for durability, not for sound improvement.
 
Re: Re: Re: Lacks a lot.

pacificblue said:

The important thing is to get the right value and the apropriate component type. Throwing money at brand names means wasting it.

Check, if the rectifier diodes are snubbered with small capacitors, varistors or RC circuits. If that is the case, forget about the Hexfreds. Else consider to snubber the rectifier and forget the Hexfreds as well.
Then again the 1N5402 is already a fast-switching diode, so you may even get away without snubbers.

If you often listen loudly, you may consider to replace the diodes with others of higher current rating. That's for durability, not for sound improvement.

The rectifier isnt snubberized and I have a freind who tried this on a gainclone and got a loss in dynamics so I wont be doing it. If Hexfred diodes are good enough for a DIY Gainclone or simmilar kit, they are good enough for my Cambridge audio. We all know faster components = better sound so why not fit the best? If you were modding a car for best performane and had serious intentions on maximising gains you wouldnt compromise on any component. You would fit the best you could get your hads on.

PS: If diodes dont affect sound quality then why do so many people upgrade them and report amazing things in terms of improved SQ?
 
Re: Re: Re: Re: Lacks a lot.

mikesnowdon said:
The rectifier isnt snubberized and I have a freind who tried this on a gainclone and got a loss in dynamics so I wont be doing it.
As I said, you might get by without snubbers. And there are reports that snubberizing can also be done wrong.

mikesnowdon said:
If Hexfred diodes are good enough for a DIY Gainclone or simmilar kit, they are good enough for my Cambridge audio. We all know faster components = better sound so why not fit the best? If you were modding a car for best performane and had serious intentions on maximising gains you wouldnt compromise on any component. You would fit the best you could get your hads on.
The best available part may not be the best part for the application. What use are the best tyres, if they won't fit on your car and don't harmonize with your chassis?

mikesnowdon said:
PS: If diodes dont affect sound quality then why do so many people upgrade them and report amazing things in terms of improved SQ?
I didn't state the opposite. My remark was only about my recommendation for higher current diodes.
 
pacificblue,

I mean no antagonism. I do get what you are saying and I appreciate that your words are sensible and considerd. I just disagree a little about the diode issue, the best diode on the market for a power amp is the Hexfred. So I'll be using them, period. I would be more in agreement with you if changing the diodes was the only thing I plan to do, then Hexfreds would be pointless. However, I plan some serious mods to this amp and I expect to get very good results. As it is with the minor preamp modifications its sounding close to its 840A brother - which uses Hexfreds by the way.
 
Administrator
Joined 2004
Paid Member
Hi Mike,
I am going to back up what Andrew and others have said. Buying name brands will certainly cost you more money for no improvement, possibly worse performance.
Mr Pass disagrees and so do many other experienced builders/designers.
...and technicians! ;)

Mike, responses like these ....
the best diode on the market for a power amp is the Hexfred. So I'll be using them, period.
will cause people who could help you to turn away and let you play on your own. If you truly want to learn something, then at least listen and learn. There is a wealth of knowledge in past threads if you wish to do some research. It's free too!

We all know faster components = better sound so why not fit the best?
? This statement has no basis in fact at all. The best component for the job has characteristics to fit the application. There are many components used in RF work (pretty fast stuff) that are ill suited for audio work. This holds true from the tube days to today's high density, high speed circuits.
However, I plan some serious mods to this amp and I expect to get very good results.
Without doing any study on your own? I doubt you will actually be successful, but you may be convinced that you are.
If you were modding a car for best performane and had serious intentions on maximising gains you wouldnt compromise on any component. You would fit the best you could get your hads on.
Now there is a semi true statement. What serious car builders don't do is design by comments off the internet or ads. They start with people having experience, add great craftsmanship and some of their own secrets they have learned over many years. Those Indy cars would be an example of this. Those are put together with closely guarded secrets. They also don't use the very best of everything. They use the best where it matters only, the other parts are good, but not that costly.

You sound like you are just starting out, with the confidence of youth. If things were truly as easy as you seem to think, then the very best audio equipment would be easily available. Everyone would be able to slap these together. Since this is not reality, what do you think the truth is?

Lastly, the "best" component depends on the exact application it is to be used in. You must know the characteristics to determine what those components are. Most component comparisons (or shootouts) are based on non-technical factors only. Humans are terrible at providing hard data. Why do you think that witnesses often counterdict each other?

Anyway, my advise to you would be to let go of the concept of "the best component". Read reports of component upgrades with great suspicion. Concentrate on component types and their attributes. You are going to have to learn something of what you are doing. The other alternative would be to go buy whatever you have read about and jam the stuff in there. Assume you have the best performance and just be happy.

How are your soldering skills?

-Chris
 
AndrewT said:
debatable!!
Mr Pass disagrees and so do many other experienced builders/designers.


anatech said:
Hi Mike,
I am going to back up what Andrew and others have said. Buying name brands will certainly cost you more money for no improvement, possibly worse performance.

...and technicians! ;)

Mike, responses like these ....

will cause people who could help you to turn away and let you play on your own. If you truly want to learn something, then at least listen and learn. There is a wealth of knowledge in past threads if you wish to do some research. It's free too!


? This statement has no basis in fact at all. The best component for the job has characteristics to fit the application. There are many components used in RF work (pretty fast stuff) that are ill suited for audio work. This holds true from the tube days to today's high density, high speed circuits.

Without doing any study on your own? I doubt you will actually be successful, but you may be convinced that you are.

Now there is a semi true statement. What serious car builders don't do is design by comments off the internet or ads. They start with people having experience, add great craftsmanship and some of their own secrets they have learned over many years. Those Indy cars would be an example of this. Those are put together with closely guarded secrets. They also don't use the very best of everything. They use the best where it matters only, the other parts are good, but not that costly.

You sound like you are just starting out, with the confidence of youth. If things were truly as easy as you seem to think, then the very best audio equipment would be easily available. Everyone would be able to slap these together. Since this is not reality, what do you think the truth is?

Lastly, the "best" component depends on the exact application it is to be used in. You must know the characteristics to determine what those components are. Most component comparisons (or shootouts) are based on non-technical factors only. Humans are terrible at providing hard data. Why do you think that witnesses often counterdict each other?

Anyway, my advise to you would be to let go of the concept of "the best component". Read reports of component upgrades with great suspicion. Concentrate on component types and their attributes. You are going to have to learn something of what you are doing. The other alternative would be to go buy whatever you have read about and jam the stuff in there. Assume you have the best performance and just be happy.

How are your soldering skills?

-Chris

Guys.

Thanks for the very good advice you have provided. It seems this thread has been a bit of a wake up call for me to some degree. Yes I am very positive about this and everything else I do. If there was such a thing as luck it could be said that its brought about by a positive attitude, and a definate persistence along a given course of action.

To get back on track: I have read very promising things about Gainclone amplifiers using the LM3886 and its derivatives. This gives me the inspiration to see what I can do with this little Cambridge audio which is allready a cracking little amp. The idea behind all this is to find out what makes a good Gainclone - good. Then to see if those factors can be introduced to the Cambridge with a little modification and research.

Its pretty widely considered that better quality componets can bring significant improvements in audio circuitry. However, I do appreciate the idea that its also important to know what type of component is suitable and more importantly how to know the correct value or rating of the copmponent. Its true that there are some of us who believe in the simple axiom of 'throw in the best components money can buy'. Certainly if done correctly this could be true. It is also true that a properly educated individual with good experience could bring about good improvenents with lesser quality parts due to knowing what the circuit is doing.

We all know that manufacturers compromise to meet the price point they wish to aim their product at. It is my belief that the blueprint of this Amp was proably too good for the targeted price range, thus compromises have been made (there are clear discrepencies in schematics). The engineers may have even deliberately 'toned down' the sound quality.

As I look around at various DIY amps and Gainclones one thing stikes me as a common truth: As we are building/modding for ourselves, and have no necccesity to make our 'product' saleable, we are able to approach it without compromise in terms of component coice. After all even the Hexfred doide is available for around £2 to £3. Panasonic FC caps are only a few quid each, etc. I have yet to see a DIY amp (or modified comercial one) that uses generic £0.50 caps or diodes in its PSU for example.

One approach I considered was to stick to the component values allready used in the amp, just replacing things for better quality items. To me that seems a safe approach. It might take a little 'component rolling' to find the right blend though. As I said this is an approach I only considered.

My point is this. I can improve this amplifier with better quality components but - as you guys have made me aware - a knowledge of which componet is suitable is tantamount to sonic improvements.

So where do I start people?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.