Return to KLUBNL.PL main page

[Top] [All Lists]

Aw: Re: [Lowfer] LF: Re: FET change, hmmm

To: [email protected]
Subject: Aw: Re: [Lowfer] LF: Re: FET change, hmmm
From: "Hans-Albrecht Haffa" <[email protected]>
Date: Fri, 30 Nov 2018 21:03:29 +0100
Importance: normal
In-reply-to: <[email protected]om>
References: <[email protected]> <[email protected]> <[email protected]> <[email protected]> <[email protected]> <[email protected]om>
Reply-to: [email protected]
Sender: [email protected]
Sensitivity: Normal
Dear OMs, dear Paul,
You are pointing to a general problem in driving fets, what is not easy, it is the same sophisticated jopb as with bipolar transistors in power electronics, in Your case a switched PA.
Please note: Any unloaded transformer winding will ring. This counts for gate drive transformers, output transformers and transformers of the power grid for 50 Hz/ 60 Hz.  In the gate drive transformer ist is possible to add damping and load resistors, but You only can damp oszillations, not neutralize the stray inductivity  of the drive transformer.  In the drain circuit oscillating may be reduced by using configurations, that are "always loaded" in some way, like in linear transistor PAs for HF. In push-pull-configuration they use a bobbin for feeding the drains and a transformer coupled by capacitors at the drains. 
To prevend current spikes by cross conduction of both fets it is necesary to use a gate drive circuit with dead time, as used in power supplies. But in this case there is a no current flow time in both transistors, what may cause spikes in the drain circuit. Damping of oscillations in the drain circuit may also reached by quiescent current, not used for class D PAs. To load the spikes to the load is not possible in an LW-PA, like in an flyback-transformer of an switch mode power supply, as the load is an LP-Filter that filters by reflections of the applied voltage.
I have failed to build a class D PA, now I have bought an industrial generator ENI 3200AMT from EGT /ENI/Electronic Industry in Rochester. I refurbished the power supply, now using three telecom PS 50 V/33 A.  Some of them are still available in USA.  It produces the usual accoustic noise and continuous power of an industrial kW-PA, but it operates...
55, Hans-Albrecht
Gesendet: Freitag, 30. November 2018 um 16:37 Uhr
Von: "Andy Talbot" <[email protected]>
An: "[email protected]" <[email protected]>
Cc: [email protected]
Betreff: Re: [Lowfer] LF: Re: FET change, hmmm
Millar feedback capacitance
The multiplication of Gate-Drain capacitance  due to negative feedback.    Don't just look at Cgs,  but also at Cgd
On Fri, 30 Nov 2018 at 15:12, N1BUG <[email protected]> wrote:
Understood but there has got to be more to it that my mind is not
absorbing. The FQA32N20C spec sheet gives 1700 pf but they are
making the waveform much softer than the FQA34N20's spec'd at 2400
pf did. What am I missing here?


On 11/30/18 9:50 AM, DAVE PICK wrote:
> Paul
> If you look at the spec sheets most of these FETs have a large
> input (gate) capacitance. Some are more than 5000pF. This is
> taking the clean edges off your drive waveform. The FQA34N20 was
> only 2450pF, so a good choice. The IRFP260 is at 4000pF, mid way.
> Seems you have to spend serious money to get the low-capacitance
> FETs!
> Dave
>> On 30 November 2018 at 14:36 N1BUG <[email protected]
>> mailto:[email protected] > wrote:
>> Thanks Dave!
>> Your note prompted me to go back to the bench and look at this
>> again. I think I just might be starting to get a clue. I'm not
>> the brightest but I really am making an effort to learn.
>> I made some more tests with the oscilloscope.
>> Looking at the output of the IR2110 FET driver *without* FETs
>> connected, the rise and fall slopes are extremely steep with a
>> clear, flat dead zone between the end of the fall on one side
>> and start of rise on the other.
>> With FETs connected but no Vdd, the slopes are far less steep
>> and the fall just reaches zero as the rise on the other side
>> begins. (with the original FETs there was still a flat dead
>> zone here)
>> Applying minimal Vdd (13V), the slopes become less steep again
>> plus they have large FET turn-on / turn-off "steps" along the
>> rise / fall slopes. Now the falling side has not reached zero
>> as the rising side is starting up.
>> I don't know how much of this is due to the speed of the FETs
>> and how much may be softening of the rise / fall due to lead
>> inductance or other phenomena in my gate circuit / driver board
>> as more current is sourced / sinked. But I do know the slopes
>> with these FETs are not as steep as with the previous FETs.
>> It should be quite interesting to install faster FETs and do
>> these checks again.
>> 73, Paul
>> On 11/30/18 8:29 AM, DAVE PICK wrote:
>>>> Paul
>>> Try the Infineon IRFP260MPBF they are fast and reasonably
>>> priced. I use Ixys IXFH58N20 but they are expensive... they
>>> are the best though! Can't find any FQA34N20 over here from
>>> reputable sources.
>>> Dave G3YXM
>>>>>> On 30 November 2018 at 12:58 N1BUG <[email protected]
>>>>>> mailto:[email protected]
>>>> mailto:[email protected] mailto:[email protected] > wrote:
>>>> Yesterday I replaced FETs in my 2200 meter amplifier.
>>>> Since the FQA34N20 is long discontinued I had no choice but
>>>> to make a substitute. I chose the FQA32N20C because I know
>>>> of one active station using them in the same amplifier with
>>>> success.
>>>> I notice drain ringing is significantly greater with the
>>>> new FETs, but maybe acceptable. I have little choice but to
>>>> run the amp and see how it goes.
>>>> I also notice the gate voltage decay takes about twice as
>>>> long as the old FETs. According to the data sheets, turn
>>>> off fall time on the 32N20C is almost twice that of the
>>>> 34N20. I suppose that is what I am seeing. I *guess* this
>>>> may be related to increased ringing, as it appears one side
>>>> is not fully off as the other begins to switch on.
>>>> I would appreciate any comments or suggestions, including
>>>> suggestions for a FET which might leave a longer dead zone
>>>> at the phase transition. I'd think slow to turn on, fast to
>>>> turn off would be nice...
>>>> Alternatively, if anyone has a stash of FQA34N20 purchased
>>>> from a well respected vendor (old stock) and would be
>>>> willing to sell some, I may be interested. I've already
>>>> tried some which originated from a Chinese source and am
>>>> not interested in going down that road again. :-)
>>>> 73, Paul N1BUG
<Prev in Thread] Current Thread [Next in Thread>