As we know all of us who mainly engaged with BJT amplifiers, the reactive circuits that protect the output stage from oscillations due to reversed currents which are produced by loudspeakers are two. First is the inductor L1 - as it appears in the picture below - that reacts in capacitive loading and second is the Zobel network constituted from R3 –C1 as it appears in figure1 that reacts in inductive loading. It should not we forget that an electric motor – such as a speaker – it can become generator. Usually the Zobel is constituted by a resistor between 5-10 Ù and a 100 nF capacitor in series as they appear in figure1. A common practice in order to see that inductor L1 reacts in capacitive loading is the parallel connection of a 2 ìF MKT type capacitor with an 8Ù dummy load at the output of an amplifier and the observation of this with the scope by applying a 10 KHz square signal in his input. Unfortunately there is not some similar method – for me at least - to observe that the Zobel network reacts in inductive loading (if somebody knows certain method it is welcome her report). Thus the only way to understand that the Zobel acts is the overheating of the resistor especially when applied square signals of 40 KHz and above in the amplifier input. Initially i used the network that appears in figure1, as many i suppose. In transient response tests of my amplifier with square signals, happened an unexpected event! The VAS transistor (an MJE15032 with Vceo = 250V / Ic = 8A) was burned. I replaced it, but afterwards burned and the new. After a lot of search, finally i discovered that was burned – something that i was afraid from the beginning because the resistor overheating - the Zobel resistor. I understood therefore how many sneaky can be this resistor if it is burned, because we are not to conceive his destruction as the amplifier continues functioning normally while he is substantially unprotected in reactive inductive loads and that consequences can result from this. Finally i led to the circuit that appears in figure2 with four resistors in parallel – series connection under the following thought. An intermediate value of 6,8Ù is satisfactory and if it happens one resistor burned the resistance will become 10,2Ù while the total power never falls under 10W. Also a MKT capacitor it is preferable than a MKP because he is faster in discharge so that it reacts better in higher frequencies and it is good to have a working voltage at 250V as reactive inductive loads usually have unexpected peaks which many times exceeds twice the amplifier supply voltage. Another item is the clamping diodes D1-D2 that withstand the Vce of Q1-Q2 to exceed more than 0,7V the supply voltage or the reverse polarity of these due to reactive loading from inductive loads - such as speakers - and it is good they bear in currents of 3A as it is MR502. I believe that it is good idea after the replacement of burned or shorted transistors in an amplifier we check also the Zobel if the damage is owed to this. I in any case, from then that applied this circuit in the output of my amplifier i resolved definitively this problem.
Attachments
full power 40 KHz sq is a bit much
for 5 W dissapation in the 10 Ohm R I get ~18 V amplitude ( 40W into 8 Ohms output load power )
no need to invoke loading effects or oscillation for the resistor burn out at higher V levels
@+/-18 V sq wave the 10 Ohm R peak pwr is ~130 W, which may vaporize the resistive element before the heat can diffuse to the body so failure in repeated surge may occur even below the 5 W average rating
the output Q destruction does sound to me more like either SOA violation from slow commutation or HF oscillation possibly after the resistor burns out
for 5 W dissapation in the 10 Ohm R I get ~18 V amplitude ( 40W into 8 Ohms output load power )
no need to invoke loading effects or oscillation for the resistor burn out at higher V levels
@+/-18 V sq wave the 10 Ohm R peak pwr is ~130 W, which may vaporize the resistive element before the heat can diffuse to the body so failure in repeated surge may occur even below the 5 W average rating
the output Q destruction does sound to me more like either SOA violation from slow commutation or HF oscillation possibly after the resistor burns out
darkfenriz said:
My friend from Poland
As i see you made a right calculation based on the equation for a first order filter Fc = 1 / 2ðRC. Even if it is somehow simplified this calculation is generally right, however it should not we forget that it loses his value in sizes that are altered unexpectedly as reported in my post. The unforeseen incident waits for always well hidden, as point out moreover big scientists – Heisenberg in “Uncertainly Principle”, Stephen Hawking in his book “A Brief History of Time” – and never can we be sure at 100% for our calculations in the paper. Simply we apply in the practice double than what we have calculated initially in order to we are then quiet. Think how easily can become antenna of reception a loudspeaker or a crossover due to inductors which contain if is found a mobile telephone near their. Our friends that are Mosfet fans know very well this. In order to verify my observation i examined enough amplifiers even without they present damage. In a percentage of 20% they presented damage in the Zobel. I do not want to frighten no one here, simply I propose - in whoever it wants - as it is careful about this item. Verify now the Zobel resistor of your amplifier. Have you lose something? As long as for my amplifier, thanks for the advice for oscillation but i forgot to report that his supply is +/- 60Vdc - individual for each channel - that means his output power vary about 200 Wrms / 8Ù ( take into account that instantly the peak power touches the 450 Wrms / 8Ù ), therefore is enough a resistor of 5W?
Dear friend jcx from the other side of Atlanticjcx said:
I forgot to report that the smaller amplifier that i have made operates with a supply of +/- 60Vdc individual for each channel. The biggest operates with +/- 86 Vdc also individual for each channel. This happens because i do not see reason for dealing with amplifiers of small power one fan of DIY. As i see you made already your calculations for an output power of 40 Wrms. And without doubt they are right. But the cost – and mainly the labour - to construct of such amplifier i believe that it touches the value of one ready in the market. On the other hand what speakers can drive this? Consequently the occupation with such power amplifiers is clearly academic - as i believe - and it is offered only as educational so that we pass then in applications of high power. I do not know for how many years you deal with audio, but me in any case roughly thirty as i am 48 years old. Therefore i am sure for a thing, that an amplifier of smaller power than a speaker can destroy it much more easily – due to clipping distortion – from an amplifier of bigger power. Big part of my experience is owed to long-lasting occupation with P.A. sound reinforcements. I am authorized service dealer of PEAVEY and CREST AUDIO in my country without this means that i copy the patents of these companies. Simply i am taught all these years from their philosophy. For this reason i have for dealing from 1985 with Mosfets and i put my constructions in so hard tests. The real life tests impose the stretching of a device under square waves with a width of 50% of supply voltage. Think now when the frequency is 20 KHz and the supply is +/- 86 Volts - hence the width of square wave is 86 Vpp - can it bears a resistor of 5W in the Zobel? If you can find an application note of Motorola (now On Semi) the AN1308 you might see in the end some oscilographs which will confirm my said. In order to i do not repeat again the same things take a look and in my answer to the Darkefenriz
Hi AndrewTAndrewT said:
Naturally and does not have relation with the Zobel network the AN1308 of Motorola. If you read my answer in JCX above, simply I recommended him to take a look in the last pages – concretely the 12 and the 15 – where they present certain test oscillograms with square waves in the input of the two amplifiers, so that he understand that is not my own opinion for how many hardly should are tested the amplifiers – specifically in figure 28, even if it is error 200KHz where is 20KHz, appear the stretching of the amplifier with square wave of 90Vpp at his output – but she is common practice in all the companies that manufacture high power devices. And this because JCX, but also Darkefenriz above, considered excessive my proposal for the strengthening of Zobel. Of course their comments were hasty because they supposed that I speak for an amplifier of 2 X 40 W/8Ù while i was reported in an amplifier of much higher power. Now why I recommended to him the AN1308 of Motorola, simply I found off-hand it front me and I know that it is free to download. Unfortunately I do not have the right to publish the service manual of high power P.A. amplifiers – professional category – of certain American companies. You would remain surprised from the self-protection systems that they contain and how much power withstands their Zobel. A company particularly uses a resistor of 40W in the Zobel! And i remained startled when I saw this for first time. However with the passage of time I became familiar with such type devices and understood very well – many times I had bitter experience – that it is not excessively what contains. Many of what I write have relation with my long-lasting occupation with such type devices, and despite what believes a lot of people they have much higher sound quality from a lot of Hi-Fi but also certain Hi–End devices. When I find time, I will publish a part from the input stage of a P.A. power amplifier that can claim very easy award regardless of that does not make the manufacturing company. They are very mystic companies; I have known this very well in seminars. Moreover they pay very well their designers in order to do not publish their secrets.
Hi again AndrewTAndrewT said:
Now want to ask you something and me. And i hope not misunderstand this. You make any construction from time to time? Or you deal academically with the object, collecting only information from all here in? Me as we say I deal not only with repairs but also with constructions for 30 years as i am 48 years old. If you have the possibility to construct an amplifier alone from the paper, send me your e-mail in eal@dra.forthnet.gr in order to post you the complete plans of my power amplifier – this is my diploma work in the Hi-Fi – and after you study it and judge that it deserves the labour make it. It is advisable because he is tried for 1,5 year of operation without any problem in the hands of other person with speakers of Mission, Spendor and Apogee.It is economic because composed absolutely from BJT (it hardly costs 350€ in Greece), and I do not want to report who known company power amplifier it had before his holder and replaced him with my amplifier. I quote a photo below. Thus precisely has his present owner him. Be quiet, I have taken all the countermeasures for the protection of his owner from electric shock!
At your service
Fotios Anagnostou
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.