Firstly I'm new here, so Hi. Secondly, I'm very new to Diy HiFi and electronics in general. However, I can follow instructions and How To's so whilst I might not understand why a circuit and component is used, I can follow an online build/instruction about what part goes where.
Anyway, a few weeks ago I stumbled across the chipamp design and started to realise that here is a good first build of a DIY amp that is both simple and cheap (compared to buying commercial). And from all accounts it sounds considerably better than a lot of commercial amps. Will it really take on and match amps costing £1000's?
Now for some more serious questions: Heat and getting rid of it. I understand the concept of heat sinks, but have no idea as to how big a heatsink to use for th Gainclone, nor how hot it will run. I've read a lot about it under normal loads it's not too bad, but what is "normal load" and what is "not too bad"?
I've also seen a lot of people use the metal casing as an extra conduit for the heat, but I'm not a fan of the normal casings you can get - unless you get something purpose built out of billet aluminium (very costly), the metal enclosures are ugly.
So that got me thinking about how I could make a case that for me is more attractive, and I've arrived at the possibility of cases out of fibreglass. I can make a mould and then lay up the glass so it's thick/strong, but I'm now concerned that the fibreglass will act too much as an insulator and trap the heat that the heatsink is giving off and so make things worse. To counter that I've come up with a plan to ventilate the top of the case via a system of "chimneys" (open tubes from the top of the case to the heat sink - I could have just drilled holes, but I wanted to make them look a bit more attractive), but again I'm not sure if this will get rid of enough heat? Finally I could drill holes into the backplate of the case (it would be hidden from view) to allow a little more air to circulate.
So, does anyone have any views on this? Ideally I want a rectangular case that is about 10 * 6 * 3.5 inches in dimension. I'd like to get the transformer into that as well, as well as a pot, but is that going to make it too crammed for a decent sized heatsink (whatever decent sized means - some help would be good here).
If I have to I'll build a seperate case for the transformer (I take it these don't get too hot as I don't see much in the way of cooling for them).
First of many questions I think. Now off to start reading my introduction to electronics book !!
Thanks for the help.
Ian
Anyway, a few weeks ago I stumbled across the chipamp design and started to realise that here is a good first build of a DIY amp that is both simple and cheap (compared to buying commercial). And from all accounts it sounds considerably better than a lot of commercial amps. Will it really take on and match amps costing £1000's?
Now for some more serious questions: Heat and getting rid of it. I understand the concept of heat sinks, but have no idea as to how big a heatsink to use for th Gainclone, nor how hot it will run. I've read a lot about it under normal loads it's not too bad, but what is "normal load" and what is "not too bad"?
I've also seen a lot of people use the metal casing as an extra conduit for the heat, but I'm not a fan of the normal casings you can get - unless you get something purpose built out of billet aluminium (very costly), the metal enclosures are ugly.
So that got me thinking about how I could make a case that for me is more attractive, and I've arrived at the possibility of cases out of fibreglass. I can make a mould and then lay up the glass so it's thick/strong, but I'm now concerned that the fibreglass will act too much as an insulator and trap the heat that the heatsink is giving off and so make things worse. To counter that I've come up with a plan to ventilate the top of the case via a system of "chimneys" (open tubes from the top of the case to the heat sink - I could have just drilled holes, but I wanted to make them look a bit more attractive), but again I'm not sure if this will get rid of enough heat? Finally I could drill holes into the backplate of the case (it would be hidden from view) to allow a little more air to circulate.
So, does anyone have any views on this? Ideally I want a rectangular case that is about 10 * 6 * 3.5 inches in dimension. I'd like to get the transformer into that as well, as well as a pot, but is that going to make it too crammed for a decent sized heatsink (whatever decent sized means - some help would be good here).
If I have to I'll build a seperate case for the transformer (I take it these don't get too hot as I don't see much in the way of cooling for them).
First of many questions I think. Now off to start reading my introduction to electronics book !!
Thanks for the help.
Ian
Welcome to the forum, and to the hobby.
Tell us what speakers you intend to drive and we can have a stab at estimating heatsink size for you. If the speakers are an easy load, you can get away with a smallish heatsink and you don't need to worry too much about the fibreglass case retaining a lot of heat. That said could you not work in some sort of ventilation grilles?
If you haven't already found it, may I suggest that you have a good look at the Gainclone pages at Decibel Dungeon, including the Gallery where you may pick up some inspiration.
Tell us what speakers you intend to drive and we can have a stab at estimating heatsink size for you. If the speakers are an easy load, you can get away with a smallish heatsink and you don't need to worry too much about the fibreglass case retaining a lot of heat. That said could you not work in some sort of ventilation grilles?
If you haven't already found it, may I suggest that you have a good look at the Gainclone pages at Decibel Dungeon, including the Gallery where you may pick up some inspiration.
Hi
Ive only been at this a week or so. Decibel Dungeon is highly recommended as a place to learn plenty Theres a good bunch of chaps here as well of course who have been very patient with my questions and of course the knowledge base herein is pretty extensive! Good luck and above all have fun
regards
Fooboo
Ive only been at this a week or so. Decibel Dungeon is highly recommended as a place to learn plenty
Theres a good bunch of chaps here as well of course who have been very patient with my questions and of course the knowledge base herein is pretty extensive! Good luck and above all have fun regards
Fooboo
Yes. Active loudspeakers designed by one of the foremost speaker experts use LM3886s. See http://www.linkwitzlab.com/ Look at the Pluto design.
Amplifiers are so good now that they're interesting only from the point of view of their cheapness or efficiency. Lots of people like designing (or buying) amps, but it's about as meaningful an exercise now as onanism.
Ah, now you're talking engineering. You can get the anticipated dissipation of your chosen chip from the datasheet, and calculate how many Watts/Degree C your heatsink needs to have given the ambient and the maximum temperature rise you want to see. Most people in amateur applications prefer to overestimate, since it's simpler and less stressful on the brain and other components.
Steer clear of unconventional materials on your first build. There's no reason you can't use fibreglass, but how many fibreglass amplifiers do you see being sold? Stick with aluminium, with wood and steel (in appropriate places) being second choices. With an aluminium chassis you can often make your heatsink integral with the box, this is a sensible use of materials.
w
Fibreglass could get a bit aromatic when warm and it's horrible to work with. If you want an illustration, try doing a clutch on a Reliant Robin, you will be scratching your arms to blood for days.
I've got LM3875s on a pair of these heatsinks in a poorly ventilated steel chassis......
http://www.madaboutsound.com/prod_det.asp?cid=87&sa=cf&ctid=7
With the volume at 9 to 10 o'clock on a 22v transformer into CSS FR125s it gets faintly warm after a couple of hours, and stays at that temperature for days. They should be well up to the job if you have suitable speakers.
Here's a non metal case design using similar heatsinks.
http://www.vikash.info/audio/tatca/images.asp
John
I've got LM3875s on a pair of these heatsinks in a poorly ventilated steel chassis......
http://www.madaboutsound.com/prod_det.asp?cid=87&sa=cf&ctid=7
With the volume at 9 to 10 o'clock on a 22v transformer into CSS FR125s it gets faintly warm after a couple of hours, and stays at that temperature for days. They should be well up to the job if you have suitable speakers.
Here's a non metal case design using similar heatsinks.
http://www.vikash.info/audio/tatca/images.asp
John
danielwritesbac said:
Yes making a chimney to encourage convection is always a good idea. Even though the fins of a CPU heatsink are too close together to be ideal, even that will work with some added convection.
Here's an example showing the apertures in the base of the case, and, in the second picture, in the top of the case with the heatsink visible.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Hi Ian,
I have a dual mono LM3875 running on a pair of 18-0-18v transformers. The chips are mounted on the rear panel (aluminium) with a couple of small (1"x 1.5") heat sinks. I play the amp through 8 ohm speakers at moderate listening levels. The amp never gets hot, and it sound great. I got the case from here http://www.hifi2000.it/default.asp?Langid=1. They were quite cheap, until sterling crashed against the euro. Very good quality, and they don't look too bad either.
I have a dual mono LM3875 running on a pair of 18-0-18v transformers. The chips are mounted on the rear panel (aluminium) with a couple of small (1"x 1.5") heat sinks. I play the amp through 8 ohm speakers at moderate listening levels. The amp never gets hot, and it sound great. I got the case from here http://www.hifi2000.it/default.asp?Langid=1. They were quite cheap, until sterling crashed against the euro. Very good quality, and they don't look too bad either.
Attachments
All
Many thanks for your welcome and response. I'm reading this in work at the moment so it's only a quick response. I'll spend more time later to look more closely at the suggestions.
Speaker wise I'll be connecting into 8 ohm Tanny F1 Customs which are only 87db efficient. I've heard that the chip amps prefer more efficient speakers so I'm guessing the Tannoys will make it work a little harder and so be hotter?
DD is a site I've already looked at many times (thanks) and I've been reading PD's thread on his chip amp build as well - I've been lurking here for the last two weeks reading as much as I can understand, but as in all things new a lot of the jargon is a bit jibberish to me, particularly all the codes for various components !!
I'll see how the Tannoy's sound, but I also have plans in my head to build some custom speakers - I've been reading up on speaker designs and came across a USA company called Audio Nirvana who make efficient drives, as well as their own speaker plans. They seem to have got some good reviews from various other forums I've been on.
I might have to rethink my design and go back to an aluminium case.
Many thanks for your welcome and response. I'm reading this in work at the moment so it's only a quick response. I'll spend more time later to look more closely at the suggestions.
Speaker wise I'll be connecting into 8 ohm Tanny F1 Customs which are only 87db efficient. I've heard that the chip amps prefer more efficient speakers so I'm guessing the Tannoys will make it work a little harder and so be hotter?
DD is a site I've already looked at many times (thanks) and I've been reading PD's thread on his chip amp build as well - I've been lurking here for the last two weeks reading as much as I can understand, but as in all things new a lot of the jargon is a bit jibberish to me, particularly all the codes for various components !!
I'll see how the Tannoy's sound, but I also have plans in my head to build some custom speakers - I've been reading up on speaker designs and came across a USA company called Audio Nirvana who make efficient drives, as well as their own speaker plans. They seem to have got some good reviews from various other forums I've been on.
I might have to rethink my design and go back to an aluminium case.
In my opinion, and regardless of factors like how loud you play, or voltage, or speaker efficiency, the 8 ohm speakers are going to help your amplifier stay far cooler than 4 ohm speakers.
One possible exception is that the crossover may or may not be hard to drive, and this is a major factor.
The "TF" versions of National Semiconductor chip amps (model code ends in "TF") are safely pre-insulated, and so you can use whatever enclosure materials you'd want to.
Given the factors mentioned, I think that a parallel amplifier with the TF chips is going to do a good job because of its enhanced compatibility with a wide variety of speakers and their various crossovers. In other words, nicely insulated (both safety and economy), it won't strain to drive speakers (it won't congest the upper mids), and that combo is one heck of a good start.
One possible exception is that the crossover may or may not be hard to drive, and this is a major factor.
The "TF" versions of National Semiconductor chip amps (model code ends in "TF") are safely pre-insulated, and so you can use whatever enclosure materials you'd want to.
Given the factors mentioned, I think that a parallel amplifier with the TF chips is going to do a good job because of its enhanced compatibility with a wide variety of speakers and their various crossovers. In other words, nicely insulated (both safety and economy), it won't strain to drive speakers (it won't congest the upper mids), and that combo is one heck of a good start.
Thanks for the continued feedback.
Sadly Maths/Physics wasn't my strongest subject so can someone give me a worked example of how to calculate a heat sinks potential heat dissipation? A Noddy explanation would be good too (treat me as if you were describing it to a kid).
As for the TF range of chips - I haven't come across them. Any pointers to somre information about them, plus designs of amps that people have built? I'm not at the stage of designing my own circuits or knowing how to modify others - I just see this as a way of spending the kind of money for an entry level consumer amp, but getting the sound qualities of a high end consumer amp.
And what do you mean by a parallel amplifier? Do you mean a monnblock for each channel powered either by seperate transformers, or one Tor with dual secondaries?
Final dumb question: Why would the insulated chips allow any material to be used? Does the insulation remove the need for heat sinks? If so, how does the chip stay cool?
Thanks
Sadly Maths/Physics wasn't my strongest subject so can someone give me a worked example of how to calculate a heat sinks potential heat dissipation? A Noddy explanation would be good too (treat me as if you were describing it to a kid).
As for the TF range of chips - I haven't come across them. Any pointers to somre information about them, plus designs of amps that people have built? I'm not at the stage of designing my own circuits or knowing how to modify others - I just see this as a way of spending the kind of money for an entry level consumer amp, but getting the sound qualities of a high end consumer amp.
And what do you mean by a parallel amplifier? Do you mean a monnblock for each channel powered either by seperate transformers, or one Tor with dual secondaries?
Final dumb question: Why would the insulated chips allow any material to be used? Does the insulation remove the need for heat sinks? If so, how does the chip stay cool?
Thanks
Insulation:
A non insulated chip delivers full voltage from one of the rails onto the heatsink. To prevent this, one searches for expensive silmic pads, purchases Kapton, lucks into finding mica, or buys the "TF" chips.
Either use a heatsink.
The PDF documents list the size in C/W as do heatsink advertisements. Search for and retrieve the Overture Design Guide spreadsheet and it will give the exact figure in C/W for a chip amplifier. That's a low tolerance figure that also assumes optimized power and so forth. Anyway, Larger is better for heatsinks.
Notice the "Tab" at the top of of "TF" chips are black (not the usual silver color). Examples: LM3875TF and LM3886TF. See the LM3875TF at www.audiosector.com
Load handling:
A parallel amplifier uses two chips per each channel, sort of like the difference in load capacity between a 1 horse carriage or a 2 horse carriage. In conditions where one would strain or start to strain (and make noise) two will not have that problem. With amplifiers, a pre-clipping sound is the "start to strain" phase. This rather irritating upper midrange errata is mostly absent in the parallel amplifier. Review (search) for spike system and general reviews of gainclones and chipamps. Notice the mention of tonality. Parallel doesn't have this problem because it doesn't strain. Sure, a big more complex on the wiring, but much less complex in the long run.
EDIT: Compared with other options (and in my opinion), you get the sound that's fun to turn up, and indeed you can do it too.
A non insulated chip delivers full voltage from one of the rails onto the heatsink. To prevent this, one searches for expensive silmic pads, purchases Kapton, lucks into finding mica, or buys the "TF" chips.
Either use a heatsink.
The PDF documents list the size in C/W as do heatsink advertisements. Search for and retrieve the Overture Design Guide spreadsheet and it will give the exact figure in C/W for a chip amplifier. That's a low tolerance figure that also assumes optimized power and so forth. Anyway, Larger is better for heatsinks.
Notice the "Tab" at the top of of "TF" chips are black (not the usual silver color). Examples: LM3875TF and LM3886TF. See the LM3875TF at www.audiosector.com
Load handling:
A parallel amplifier uses two chips per each channel, sort of like the difference in load capacity between a 1 horse carriage or a 2 horse carriage.
In conditions where one would strain or start to strain (and make noise) two will not have that problem. With amplifiers, a pre-clipping sound is the "start to strain" phase. This rather irritating upper midrange errata is mostly absent in the parallel amplifier. Review (search) for spike system and general reviews of gainclones and chipamps. Notice the mention of tonality. Parallel doesn't have this problem because it doesn't strain. Sure, a big more complex on the wiring, but much less complex in the long run. EDIT: Compared with other options (and in my opinion), you get the sound that's fun to turn up, and indeed you can do it too.
westers151 said:
There's an excellent question within that question! I'm so sorry that I almost missed it.
Dual Mono is, electrically, a pair of monoblocs that happen to be within a single enclosure and use a single stereo volume knob.
With the gainclone amplifiers, the term meaning "in the style of 47 labs, with low parts count," all of the parts are carefully chosen.
An example of a Dual Mono parallel LM3875TF gainclone:
One enclosure
One stereo volume control
Two transformers (one per "right" channel, one per "left" channel)
Two rectifer boards (one per "right" channel, one per "left" channel)
Four amplifier boards (two per "right" channel, two per "left" channel)
Parallel is more complex and more expensive. It is an unnecessary approach. I mention it because it has a laid back sound due to the fact that it does not strain. Directly after building the amplifier, you'll be less likely to have (typical of gainclone) questions about tone controls because you won't need them. See? That's easier, I think.
OK, a simple example might be a heatsink for a linear regulator - LM7805CT. Suppose the supply voltage is 12V and the current in the circuit supplied by the regulator is 1A (the max).
The power being dissipated by the regulator is 7V * 1A = 7W. this is the voltage being dropped by the regulator:- 12 (the supply) - 5 (the output voltage) = 7V times the current in the circuit.
Looking at the chip datasheet, the thermal resistance of the device (junction-air) in a TO220 package is 65 degrees/Watt. In this case a temperature rise of 7*65 = 455 degrees. Add this to an optimistic ambient of 30 degrees C, and you have a junction temperature of 485 degrees C, which is clearly unacceptable.
Returning to the datasheet, the junction-case thermal resistance is 5 degrees/Watt. If we take a more conservative ambient, say 40 degrees, and say that we want a maximum junction temperature of 100 degrees (max on the datasheet = 125) now we have 60 degrees of permitted temperature rise to be generated by 7 Watts. This requires a total thermal resistance of of 60/7 = ~8.6 degrees/Watt. Since the junction-case resistance is given as 5 degrees/Watt the heatsink must be 8.6-5 = 3.6 degrees/Watt or better (smaller number). That presumes no penalty for the package-heatsink junction. Resistances in series are additive.
I've chosen a regulator for the example since I can arrange for the dissipation to be a convenient constant. This is not necessarily the case for an amplifier, and determining the true dissipation may be less than straightforward, but I ain't gettin' into that... anyway, as already stated, the datasheets for most of these chips have specific recommendations.
w
The power being dissipated by the regulator is 7V * 1A = 7W. this is the voltage being dropped by the regulator:- 12 (the supply) - 5 (the output voltage) = 7V times the current in the circuit.
Looking at the chip datasheet, the thermal resistance of the device (junction-air) in a TO220 package is 65 degrees/Watt. In this case a temperature rise of 7*65 = 455 degrees. Add this to an optimistic ambient of 30 degrees C, and you have a junction temperature of 485 degrees C, which is clearly unacceptable.
Returning to the datasheet, the junction-case thermal resistance is 5 degrees/Watt. If we take a more conservative ambient, say 40 degrees, and say that we want a maximum junction temperature of 100 degrees (max on the datasheet = 125) now we have 60 degrees of permitted temperature rise to be generated by 7 Watts. This requires a total thermal resistance of of 60/7 = ~8.6 degrees/Watt. Since the junction-case resistance is given as 5 degrees/Watt the heatsink must be 8.6-5 = 3.6 degrees/Watt or better (smaller number). That presumes no penalty for the package-heatsink junction. Resistances in series are additive.
I've chosen a regulator for the example since I can arrange for the dissipation to be a convenient constant. This is not necessarily the case for an amplifier, and determining the true dissipation may be less than straightforward, but I ain't gettin' into that... anyway, as already stated, the datasheets for most of these chips have specific recommendations.
w
Daniel/Waki,
Many, many thanks for the input - you're filling in the basic "jargon" that's so critical to understand any subject. Without your input I would still be very confused. Now I'm merely confused
I know I'm a "newbie" at this, but I'm interested in doing a dual chip per channel. I'm sure some people will say "walk before you can run" with regards to building, but I'm not too fazed by building something if there's a manual/guide that can help me do it. After all, it's a bit like painting by numbers - anyone can do it if they can a) read, b)translate the number into the correct colour, c)paint within the lines.
Lots to think about, but I feel I'm now grasping some of the basics, even if there are a lot of basics still to be understood.
Ian.
Many, many thanks for the input - you're filling in the basic "jargon" that's so critical to understand any subject. Without your input I would still be very confused. Now I'm merely confused
I know I'm a "newbie" at this, but I'm interested in doing a dual chip per channel. I'm sure some people will say "walk before you can run" with regards to building, but I'm not too fazed by building something if there's a manual/guide that can help me do it. After all, it's a bit like painting by numbers - anyone can do it if they can a) read, b)translate the number into the correct colour, c)paint within the lines.
Lots to think about, but I feel I'm now grasping some of the basics, even if there are a lot of basics still to be understood.
Ian.
Well, have a trip over to www.audiosector.com Its a good read.
Ooh, also Decibel Dungeon, located at: http://myweb.tiscali.co.uk/nuukspot/decdun/gaincloneindex.html
Ooh, also Decibel Dungeon, located at: http://myweb.tiscali.co.uk/nuukspot/decdun/gaincloneindex.html
westers151 said:
There's a walk before you run approach available and quite useful. Its the Audiosector kits.
The "sticky" thread at the top of the chipamp forum has "Commercial gainclone building. . ." which is the Audiosector gainclone amplifier assembly guide and support thread.
Since the Audiosector parts match so exactly, you could start with the ordinary Stereo model (one transformer, one rectifier board, two amplifier boards) and then later expand this if you wish.
Notes. . .
A caveat applies to gainclone style: The low component count amplifiers are highly reliant on a careful choice of components.
You'll need the exact model capacitors specified because they were carefully chosen and you don't need to repeat all that work unnecessarily.
You'll need the type (Toroid), Voltage, and VA size transformer specified.
Expansion options?: You can use parallel if you wish and/or you can use a buffer. See Decibel dungeon about that buffer. It is perhaps a good option for tonality that is less expensive and possibly even better than a parallel amp with no buffer.
See? It doesn't have to be expensive.In this, a well-done Audiosector basic LM3875TF stereo amplifier is a reference point. You can use this reference point while building a lovely buffer. See also Rod Elliot's method for hooking up headphones because this is much more useful than speakers when judging the success of an amplifier and/or buffer.
Ok, can I just check that I've understood the heatsink thermal dissipation calculations.
LM3875 spec sheet says:
If I assume an ambient temperature of 40c, decide that I want a Max Junc rating of 50c (just in case someone grabs the case I'd rather have the temperature down low to stop burns), then my calculations give me the following:
So I need a heatsink that has a rating of 0.92 c/w or lower.
Correct?
So if I find one that is rated at 0.4 c/w then there's a fair chance the heatsink will hardly get warm.
LM3875 spec sheet says:
- Junction-Case rating of 1 c/w
- Max Junction rating of 150c
- 125w dissiption
If I assume an ambient temperature of 40c, decide that I want a Max Junc rating of 50c (just in case someone grabs the case I'd rather have the temperature down low to stop burns), then my calculations give me the following:
- 50c - 40c = 10c permitted rise
- 10c/125w = 0.08 degrees/Watt
- 1 c/w - 0.08 c/w = 0.92 c/w
So I need a heatsink that has a rating of 0.92 c/w or lower.
Correct?
So if I find one that is rated at 0.4 c/w then there's a fair chance the heatsink will hardly get warm.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.