Selasa, 15 Desember 2009

CCS part 2

Sebelum dilanjutken, perlu digaris bawahi bahwa tulisan ini merupakan kumpulan diskusi audio designer disebuah group ( lupa namanya ) sekitar thn 2004-2005.
Saya mah cuma alumni SD INPRES, ... nggak nyampe otaknya.
Jadi kalo bingung jangan tanya sama saya yah...
Judul diskusi wis kelalen, mungkin CCS vs Choke ?

Detailed characteristics of a CCS:
1. A CCS can not swing above B+.

1.1 A CCS has a minimum voltage that can appear across it for it to
function or it will clip like an amp hitting the voltage rail. This minimum voltage can vary from a couple volts or tens of volts. For the coming example, lets call it 18 volts. This 18 volts has to be added to B+.

1.2 Half the peak to peak AC swing of the tube also needs to be added to
B+ to keep the CCS from clipping. (I like to add this voltage in assuming the tube is unloaded.) So if you are getting 400V pp out of the tube, B+ goes up another 200V for a total of 218V.

1.3 The B+ now has to have margin in it for line voltage variation. If B+ drops 50V because of low ac line, that is 50V less peak voltage the CCS can deliver, meaning the CCS will clip at a lower power out. This can be improved by using and "active load" CCS or cured by adding margin to B+ to cover line variations.

2. A CCS does not have a DCR. The closest thing to DCR is the minimum voltage drop needed for the CCS to work.

3. A CCS has an AC parallel resistance that is fairly flat with voltage
and frequency and is usually many times higher than the parallel resistance of an inductor. This AC resistance can be from 100's of kohm to >10,000 megohms.

4. The AC parallel capacitance of a CCS can be the same as an inductor or
so small the 6 inch wire to the CCS has more capacitance.

5. A CCS does not really have inductance. If it did, it would be infinite (megahenries). The effective impedance of some CCS will vary with voltage, but seldom with frequency. The amount depends on the design of the CCS.

6. A CCS delivers the same current all the time. That's why it is a CCS. A CCS can be designed to have a very slow turn on of the delivered current. This is something you can't do with an inductor.

7. CCS usually need heat sinking. NEVER believe the marketing rating for the power of a solid state CCS part. Transistor's power ratings are specified with the transistor attached to chilled, water cooled heat sinks that hold the surface of the transistor at 25C no matter how many watts are lost in the part.

7.1 Here's a problem with heat sinking CCS's. Many metal tab transistor
will arc over to the mounting screws (even with good insulating bushings and thermal mounting insulators) in the low hundreds of volts. If there is a burr on the heat sink, all bets are off. That burr is a failure just waiting to happen. (Don't use the blue silicon insulators, they tear so easily it isn't funny.) Use TO-220 fullpaks or TO-247 packages, these failure mostly go away.

7.2 Modern switch mode FETs can't be used in the linear region anywhere
near rated power. They internally have problems with current hogging. Fairchild Semiconductor has a article about this that I can't find back. Buy your
CCS FET based on smallest drain capacitance, highest voltage rating and in a minor part, the smallest Rtheta junction to case. This usually happens in the highest Rds-on part (4 ohm), not the lowest (0.8 ohm).

7.3 CCS usually have a "safe operating area" (SOA) that has to be watched. Bipolar devices have more SOA problems than FETs, but Bipolar devices have lower drain capacitance than FETs.

7.4 Remember, Silicon does not have a sense of humor. When it says 300V rating. At 301V peak instantaneous, the part can violently die. (The parts I have measured from on-semi had 15 to 20% margin at room temp.) When a tube says 300V rating, it means the DC bias point can be 300V and the ac peaks can go higher.

8. Most CCSs have very little or no magnetic pick up. Yes I can design one that would, but I'd have to try to do it.

9. Most CCS can have some capacitive coupling issues, but it is easy to
make the parasitic capacitance of a CCS 100 times or more better than an inductor.

10. Most CCS won't be sensitive to vibration, but with the wrong choice in parts they can be sensitive to vibration.

11. A well designed CCS will have an excellent transient and overload response. A poorly designed one won't. A CCS is an amplifier circuit in it's own right. Good ones make the music made by the amp sing, bad ones make it belch.

12. CCS can have current distortion. When measured in uA, you'd have to
Try to design one didn't work significantly better than an inductor. When measured in THD, they don't look as good. That is because the fundamental current change is almost zero! Example: 1 uA distortion (50%) on a 2 uA fundamental is bad when it comes to THD, but it works better than 150 uA (3%) distortion on a 5 mA fundamental.

Other notes using CCS

A. For a simple tube, you can CCS the plate or CCS the cathode. You can't CCS both. It is impossible to exactly match CCSs. So if you put a 1.0000 mA CCS on the plate and a 1.0001 mA CCS on the cathode, the bias point won't be right.

B. On a long tail pair, you can CCS two of the three connection, but not all three. You can CCS both plates or CCS one plate and the tail or just CCS the tail.

Warning. . .Inductor Distortion Soap Box:
The current distortion from an inductor must be specified in uA at each harmonic with a given drive voltage and frequency, not in THD. Most of the inductor's distortion is from the iron, not the air gap or other effects.
Increasing the air gap and keeping the copper the same (which drops the inductance) will make the THD look better, but the uA of third harmonic distortion does not significantly change. This is because the fundamental current flow has increased and the distortion components stayed about the same. The fallacy here is the fundamental current has increased causing the tube to makes more distortion because the load line is more elliptic etc but the tube still has the same amount of distortion current flowing in it from the inductor as it did before.

Now if the copper is redesigned so the inductance (and core material and size stays the same) but with a larger air gap, now the lower THD is good because the harmonic distortion in uA is now lower. The penalty for this change is a higher DCR and higher temperature rise.

== end of discussion ====

Tidak ada komentar: