DIY books

[hacknet 0]

sorry for being a wet blanket but why dB the thing?

 

the way it looks, it seems like a rather crude DB with tons of diodes used for biasing.. i think you would not achieve as good distortion figures as even a simple bufXXX IC.

 

the reason i see that most people employ buffers is because there is a lack in current drive. there is a simple and effective way to solve this.

 

stack opamps.

 

for example stacking maby 8 opamps in parrallel. this would improve current drive, improve linearity, increase the damping factor and all most of all, it certainly distorts less than a DB without proper biasing.

 

inside an opamp, there are temperature and current degredation compensation mechanisms, output devices are perfectly matched and the list goes on. one thing a DB would certainly loose out to an opamp would be its inability to serve a wide bandwith. DBs have very limited bandwith and slewrate even compared to a 5532 as chunks of capacitance wait along the wire links.

 

i problem i see with the design above, there is no DC offset compensation pot. you are probably going to find DC on the output unless you very carefully match all the components including the drivers and its surrounding resistors.

 

anyway. good luck!

[fwh1599 0]

I think use DB as output stage is very simple and good,we can use BUF IC or discrete DB.

As it:

Edited May 7, 2004 by fwh1599

[jtfoo 0]

Have you stacked opamps before?

 

Opamps don't usually behave well when stacked,especially when you have a feedback loop. They'll fight each other and burn out.

 

It's still possible to parallel them if you use series resistor at the output. Some tried and succeed, but still many headamps diyers don't stacked opamps

 

[fwh1599 0]

don`t mind me asking but why do you need a discrete stage?

 

discrete stage tend to add noise, distortion and drift since they are not perfectly matched and the wiring is long compared to the tracks of a silicon die.

 

is there any particular reason you choose to add a discrete stage?

hacknet:

I think use a discrete output stage can increase the Amp's drive capacity,and the AMP is a loop feedback,if select low offset and JFET input OP,you can obtain very low offset.

In my twinkle(My Twinkle),I use AD8065,the offset is less than 1mV,if use NE5534 the offset is about 10mv.

 

jasonhanjk:

I think A1015/C1815 isn't fit output stage,it's a TO-92,Pc<=400mW,current <10mA(+-24V supply,no heat sink),along with the use of D7-D9 and emitter-stabilization resistors R8-R9,so R8=R9>100Ohm.In my Case,I use MJ15030/15031 at 60mA(emitter-stabilization resistors =10ohm),I can get low distortion,Class A sounds!

Edited May 7, 2004 by fwh1599

[hacknet 0]

Have you stacked opamps before?

 

Opamps don't usually behave well when stacked,especially when you have a feedback loop. They'll fight each other and burn out.

 

It's still possible to parallel them if you use series resistor at the output. Some tried and succeed, but still many headamps diyers don't stacked opamps

yes, i have heard stacked opamps before. they sound good.

 

it is unlikely they fight each other especially if they are from the same batch...

 

 

>fwh1599> well, opamp`s offset is one thing, the output stage`s offset would be another. if you do not match the devices properly, you will end up with DC on the output because the 2 device are not complientary. alot of care has to be taken to ensure that matching is done properly. even if it is done properly, there is no temperature compensation(unlike an opamp) to correct the drift when the devices get hot. after a couple hours of operation, you might end up with DC even with matched devices.

 

more over, with a push pull topo, you have the zero crossing problem. only way to solve this is to raise the bias. higher bias = more heat = more drift = more distortion.

[jasonhanjk 0]

Thanks guys for your valueable inputs. I use 1015 and 1815 because got lot's of em (around 50 each). Just got another drill from KaiChin...

[jtfoo 0]

yes, i have heard stacked opamps before. they sound good.

 

it is unlikely they fight each other especially if they are from the same batch...

 

 

hacknet,

 

If you recommend stacking opamps, which btw many people don't, it helps to put some word of cautions.

 

. Because you have to get matched opamps

. Unlike mosfet which can be matched via measuring Vgs, how do you matched opamps?

. Even on the same batch, chances of mismatched are still high.

. Stacking two opamps is already a gamble, now what's the probability of having 8 matched opamps, the number you recommend?

. It all boils down to predictability. Stacking two opamps don't guarantee it work, it just luck, taking from the same batch just increase your odds. Unlike buffers (e.d EL2001) where you take any good one and they can be stacked in numbers irregardless of batch, predictable results. Parallel opamps with output resistors are predictable.

. Result of opamps fighting each other can lead to opamp burn out. Damage opamps, very expensive if all 8 goes.

. And your disclaimer of 'try at your own risk, not responsible for any damaged'.

 

Circuit with stacking opamps, haven't seen one yet, probably you can share with us your circuit.

Parallel opamps, I seen a few and even on some application notes.

Parallel with master slave config, the Apheared47.

Edited May 7, 2004 by jtfoo

[fwh1599 0]

hacknet:

 

I get low offset at output stage,I hear Twinkle about a month.

 

I think mabey these are important for get low offset:

1.Select lower offset OP;

2.Select higher PSRR OP;

3.Very symmetrical +-supply;

4.Matched component(Transistors,Diodes,Resistance).

[hacknet 0]

i`ve seen it done before....

 

 

go try it and let me know if it works in your case...