Hybrid amplifier.. Please help..

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hey everyone,
I posted about this in the Tube section a while ago but i thought i should post it here now because it is related to the SS section of the amp.
I am in the process of putting together the first prototype of the following on a breadboard, any reasons why it wont work? anything i should change before i power it up?
I know i may have problems with DC offset but there will be an output capacitor in it for testing and i'll sort that out later after i know it works or not.
Anything else i have missed?
Power supply will be +/- 30V
MAny thanks,
Owen
 

Attachments

  • hybrid2ccs.jpg
    hybrid2ccs.jpg
    32.1 KB · Views: 1,242
Your R5, VR1 and R8 need to be proportionally bigger. As they are now, they will be passing most of the current instead of the BD139 used as the bias servo. This will make the BD139 less effective than it should be.
Try R5=10k, VR1=4.7k, R8=2k7. You have approximately 4x transistor Vbe from one input to the other. This is the bias voltage of your output stage. The circuit with the BD139 is called a Vbe multiplier, and it multiplies the Vbe of the BD139 by 1+(R5/(VR1+R8)). For the lowest setting of VR1 (=0 ohms), look for R5/R8 to be slightly more than 3, altogether this will guve you a bit more than 4x Vbe for maximum bias, and still you will be able to get bias to zero with the VR1, for initial testing.
You can use one cap across inputs IN ADDITION to the two caps that couple the SS part to your tube front end. In most designs the third cap is not used as the two input caps essentially double up as the third one, just of a much lower value, sometimes too low. In these cases, the third cap like AKSA mentiones may prove useful.
 
so if i use say.. a 2.2uF polyprop cap on each from the tube stage to each of the inputs and have a 22uF non polar electrolytic cap bypassed with a o.1uF poly cap accross the two inputs this would be a better way of doing it? :) Thanks for the advice on the resistors, i will try what you suggested, i take it to get the bias to 0 i have to have VR1 set to its maximum?
this is my first real jump into SS, i have built quite a few tube circuits before and have a HND in electronics engineering, i just am not really familiar with SS audio circuits yet.. but i have learnt a lot in the last few weeks.. :)
Many thanks,
Owen
 
Your C1 is not necessary if you already have C4, and C5 is not necessary if you have your two 'IN' points connected together before the 2.2uF caps (like they should be), as the two 2.2u caps in series are already in parallel with C4.
You also need about 100n in parallel with R3, like AKSA suggested.
Ultimately, Sajti is correct, it will not work without making a potentially huge DC error on the output, which will make it behave completely differently than it's supposed to, assuming it does not, by some miracle, stick to one rail. I'm sorry if this is blunt, but for someone with a HND in electronics engineering, I find it a bit odd that in this thread, people have been correcting errors that reflect unfamiliarity with basic electrical rules.
Without a load, your circuit will saturate to one rail, because of current source inequality (that would be your R4, R7, D1, D2, Q6 and presumeably Q7). You have two current sources connected in series - the voltage at the connecting node is undefined for the ideal current source, unless there is an impedance from that node to a voltage rail or ground, to close the current loop and define the voltage by the drop across it, with the difference of currents of the current sources, passing through it.
For real current sources, which are non-ideal, the connection node will be close to either of the rails. The circuits around it are a voltage follower, so the output voltage will, as the name suggests, follow - without a load, or with an AC coupled load, it will stick to one rail. With a load, one uA of input current difference, will easily make about 10mA of output current difference, i.e. 80mV across 8 ohms of load. Except that you will be VERY lucky if your current sources are that close, count on 100s of times the error above.
 
Keep it simple.
Why not get rid of the current sources altogether and replace them both with 10k resistors? This will keep the offset manageable and will be fine provided the driving stage can drive 5k ok.
I would advise against a cap across R3.
With the 2.2uF polys connected together you really don't need C4 or C5 and a half-decent 10uF for C1 as you had will be just dandy (don't guild the lilly with exotic capacitors at this stage) - your silicon is not worthy.

As you are new to silicon be aware of a few things. Transistors are extremely non-linear and have much reduced bandwidths compared with valves. Driving them requires much consideration as they have realtively low and non-linear impedances. Complimentary pairs are usually not, at least not enough for audio.
 
traderbam said:
Keep it simple.
Why not get rid of the current sources altogether and replace them both with 10k resistors? This will keep the offset manageable and will be fine provided the driving stage can drive 5k ok.
I would advise against a cap across R3.
With the 2.2uF polys connected together you really don't need C4 or C5 and a half-decent 10uF for C1 as you had will be just dandy (don't guild the lilly with exotic capacitors at this stage) - your silicon is not worthy.

As you are new to silicon be aware of a few things. Transistors are extremely non-linear and have much reduced bandwidths compared with valves. Driving them requires much consideration as they have realtively low and non-linear impedances. Complimentary pairs are usually not, at least not enough for audio.


And if You use 4x5kohms instead of 10kohms, You can do bootstrapping from the output to the common points of the resistor pairs, to get very high input impedance without complicated DC servo circuit.

Sajti
 
my original circuit DID use resistors instead of current sources, that's how i was told to set the current sources up, by someone on these forums, not only to help improve stability but also to increase the input impedance slightly because the tube stage was haveing problems running it...
So Should i Go back to useing the resistors instead of the CCS and do as sajti says and use bootstrapping?
atached is my ORIGINAL design... that is the one that i posted on here a month or so back..
Would i be better off just changeing some values on this one and adding the bootstrapping? I take it for the bootstrapping i use 2 x 5K resistors in series instead of the current source and sink and connect the midpoint of each of these two resistors to the output?
many thanks,
Owen
 

Attachments

  • hybrid2.jpg
    hybrid2.jpg
    27 KB · Views: 431
Administrator
Joined 2004
Paid Member
Hi Guys,
I've build similar circuits. I told Owen to go to current sources for a couple reasons. Supply rejection and input impedance are the two main ones. Practically, this solved some problems for me.

DC offset.
No servo required. If you get you input impedance up some more (another stage) it's very easy to reference each drive point to ground through a pair of resistors. Mine are max 40mV from ground, although my circuit is very slightly different. If you really need to add a servo, just drive this common point for DC correction.

You have to remember that Owen is planning to drive this from a tube stage. A bipolar output stage is vastly more linear than a mosfet output stage.

Owen,
You need to use those signal transistors you ordered for the current sources. The BD parts will work, but the others will do a much better job.

-Chris
 
hey chris,
I thought you might reply here, I think i am going to test the circuit with and without the current sources and see which works best for me, The power supply i have made for it has very little ripple on it (surpriseing really as it is very basic)

The real Question is.. Which will work best in this situation... I like the idea of useing resistors instead of the current sources because it cuts down on the active devices in the circuit..
Owen
 
Administrator
Joined 2004
Paid Member
Hi Owen,
I don't have a problem with you testing both ways .. I encourage you to do this. You may find your 2 nd harmonics go way up at higher levels with the resistors.

Using resistors also makes the referencing to ground more difficult as the impedance is much lower. This alone may force you to use more active circuit elements in the form of a servo.

Remember that your tubes want to "see" a very high impedance load or the total distortion will be dominated by the tube stage.

-Chris
 
Hi chris,
So useing the resistors and bootstrapping to the output will show a lower input impedance than useign the current sources?
This is all not really important at the moment anyway, i'm pretty sure i am going to have to add an extra stage liek we discussed before to in order to sucessfully drive it from the tube abyway.
I have started to put it together on a breadboard without the current sources, it will be easy to add them later anyway.
Many thanks,
Owwn
 
Bootstrap 'current sources' are only as good as their capacitors and the amplifiers performing the bootstrap, in your case, your SS output stage. They 'multiply' the resistor between the cap joining node and the input by 1/(1-Av), Av being the voltage amplification of your SS output stage, which is close to 1. Therefore, the bootstrap resistor is multiplied many times. Even so, effective Av may reduce due to various reasons, or, which is the usual problem with complementary followers, become nonlinear (especially for low bias currents). Because of this it may be prudent to increase the bootstrap resistors as much as possible in your circuit, as your tube front end wants to see a high impedance, the higher the better.
To keep things the same as in the CCS version, you want the same current going through your bootstrap chain as would be through the CCS. The voltage span from the rail to the input is about 20V - (2 x Vbe) so let's say about 28.6V. Your CCS will provide about 1mA (depending on the LED type), so for 1mA across 28.6V (keeping things equal here), you need a resistance of 28.6k. Closest would be 27k. However, for a bootstrap arangement, you want to split that into two resistors. My advice would be 12k each. So, liek this:
+ Supply rail - 12k - 12k - upper input, connect output to joining point of the two 12k with bootstrap capacitor (say 100uF).
- Supply rail, 10k - 4.7k trimmer - 12k - lower input. Connect bootstrap cap between joining point of 12k and trimmer, and output (same 100uF).
This gives you two bootstrap 'current sources' at ~1mA, and a means to adjust offset both into + and -.
With proper bias of the output stage, bootstrapping action will easily increase the input impedance (without bootstrapping it would be approx. 12k) 50-fold at worst, which should be just fine for your tube stage.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.