Message-ID: <5C7409C9.9000803@posteo.de>
Date: Mon, 25 Feb 2019 16:29:13 +0100
From: DK7FC <selberdenken@posteo.de>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <043bec62-8144-5ff3-98a6-9f04699f6a4c@no3m.net>,<5C73F86B.8080201@posteo.de> <1551107358099.3307@kuleuven.be> <5C740794.1040902@posteo.de>
In-Reply-To: <5C740794.1040902@posteo.de>
Subject: Re: LF: Class D current spikes
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

PS: In a current mode call-D PA i would rather expect an overvoltage 
issue rather than an overcurrent issue (at the power supply). Maybe the 
sudden stop of the current consumption (100% to 0%) is to fast for the 
regulator of the SMPS, so it does not regulate the duty cycle fast 
enough and produces its own overvoltage. Or, a voltage peak across the 
terminals due to the high di/dt across the L of the cable between SMPS 
and PA.
Such problems could be minimised easily by some additional C near the 
SMPS...

73, Stefan

Am 25.02.2019 16:19, schrieb DK7FC:
> ...yes, might.
> But also it might not, especially when the critical event happens only 
> once per minute.
>
> 73, Stefan
>
> Am 25.02.2019 16:09, schrieb Rik Strobbe:
>> Hi Stefan, Eric,
>>
>> this might solve the PS tripping problem, but on the other hand it 
>> might also provide the MOSFET the draincurrent it needs to die ;-)
>>
>> 73, Rik  ON7YD - OR7T
>>
>> ________________________________________
>> Van: 
>> owner-rsgb_lf_group@blacksheep.org<owner-rsgb_lf_group@blacksheep.org>  
>> namens DK7FC<selberdenken@posteo.de>
>> Verzonden: maandag 25 februari 2019 15:15
>> Aan: rsgb_lf_group@blacksheep.org
>> Onderwerp: Re: LF: Class D current spikes
>>
>> Hi Eric,
>>
>> What happens to the supply behaviour if you just switch some Cs parallel
>> to the output, say 4x 4700 uF (low ESR) rated at 63V ?
>> Maybe it solves the problem of the supply going into fault mode,
>> although it is no the perfect solution.
>>
>> 73, Stefan
>>
>> Am 24.02.2019 13:08, schrieb Eric NO3M:
>>> I recently got a current mode (outp xfmr center tap not bypassed)
>>> Class D amp going on 136khz, basically modeled after the W1VD / G0MRF
>>> circuits.  FETs are 59N25s, 4820p / 10ohm snubbers, 5t+5t / 21t 3C90
>>> 42mm output xfmr, 62uH (25t T157-26) choke.  Output xfmr depends on
>>> the power supply in use:
>>>
>>> 30V linear supply: 5t+5t / 21t 3C90 42mm, 300W out
>>> 48V switching supply: 5t + 5t / 16t 3C90 51mm, 500W out
>>>
>>> I am also using the G0MRF keying circuit, ie. P-channel MOSFET
>>> (IRF9540) in the driver (MCP1404) supply line.
>>>
>>> If the driver is left active (key down), the beginning and end of the
>>> input RF envelope produces a "clicking" noise... assume this is from
>>> current spikes being developed and sunk by the FET body diodes?  If
>>> the driver is keyed with input RF already applied, thereby
>>> interrupting the drive to the FETs, there is no such behavior.  The
>>> keying is shaped, resulting in a ramping of the driver supply voltage
>>> and resultant gate waveforms to the FETs.
>>>
>>> These current spikes do not bother my 30V linear power supply,
>>> however, when I attempt to use a 48V switching supply (rated at
>>> 17.7A), it seems to be OK with the current spike generated on the
>>> beginning of the RF envelope, but not the spike at the end of the
>>> envelope and goes into fault mode.  If the driver is keyed with
>>> continuous RF applied, it's happy and does not trip out.  But keying
>>> the driver doesn't seem practical for digi modes without some exotic
>>> circuitry or engaging computer generated signaling to key down *after*
>>> RF drive is applied and key up *before* RF drive is removed.... again,
>>> this phenomena only happens at the beginning and end of the input RF
>>> envelope.
>>>
>>> Is there way to mitigate these current spikes, especially the one
>>> generated when RF is removed?
>>>
>>> 73 Eric NO3M
>>>
>>>
>>