
4/6
www.dynexsemi.com
AN4999 Application Note
Cathode
Anode
Main
thyristor
Pilot
thyristor
Gate
The other extreme condition is where one thyristor in the
paralleled batch has an exceptionally low spike voltage. This
device turns on early and can hog all the current.
Generally, the lower the supply voltage the worse are the
paralleling problems.
In practice, a small increase in busbar inductance in series with
each thyristor can usually overcome the turn-on problem.
DEVICE SELECTION
How can devices be selected to overcome paralleling problems
The easy solution is to include large series reactors so that mis-
sharing is negligable, even with unselected devices.
If this is not possible the following procedures should be adopted:
Maximum effort should be made to equalise busbar lengths.
This will help both steady state and turn-on sharing.
Ensure that as near as possible the device case temperatures
are equalised.
Ensure that a high gate drive current is used (similar to that
recommended for good di/dt rating).
Finally, determine the equipment operating current at which
current sharing is most important. Usually, this is the full load
operating current but sometimes overload current is more
important.
At this stage a preliminary choice of thyristor should be made
for continuous operation, assuming a mis-sharing of current
between devices of, say, 15%:
Define mis-sharing factor, ‘m’ as:
m = (I
max
- I
min
)
x
100%
I
max
Where I
is the current flowing in the highest current device
and I
min
is the current flowing in the lowest current device.
From the thyristor supplier determine the available V
f
selections for paralleling at the T
j
and current you require.
Calculation procedures can then be used which input device
maximum and minimum V
curves and the device busbar
connection inductance to determine steady state mis-sharing
factor.
Finally consider turn-on mis-sharing due to turn-on spike
voltage variations. This is unusual unless very short busbar
lengths are used. Also, the problem only affects a few device
types but if a problem is likely it should be mentioned to the
supplier in case a selection is needed.
This corresponds to about 3V anode to cathode which is the
minimum needed for turn on of a large thyristor. Clearly, a gate
pulse width of 10 to 15us is OK for a switched load of several
hundred volts but a 1ms pulse is needed for a 10V source.
Unequal conduction due to transient on-state voltage
variations:
As indicated above, thyristors need a minimum anode to cathode
voltage to turn on. This is indicated by the well known thyristor
switching characteristic, Fig 5. The minimum value is sometimes
known as the
‘
finger voltage
’
It is clear that a group of thyristors connected in parallel with
different finger voltages and different delay times could present
a problem for turning on.
The problem is sometimes worse when amplifying gate devices
are used. The amplifying gate thyristor is effectively a main
thyristor, darlington driven by a pilot thyristor connected between
anode and gate, Fig 6.
When an amplifying gate thyristor is triggered current initially
flows in the pilot thyristor until enough current is flowing to trigger
the main device. Then the circuit current transfers to the main
thyristor. Usually, the initial voltage across the pilot thyristor must
reach a minimum level to initiate transfer. This minimum level
appears as a
‘
spike
’
on the Vf waveform. Fig 7 is an example of
a high spike voltage.
‘
Spike
’
voltage values can vary across a
batch of devices.
In a paralleled system, if one of the thyristors has a high spike
voltage it may be late to turn on and so conduct for only part of
the half sine wave period. The consequence is a low average
current. In the worse case the device may fail altogether to turn
on.
Fig. 6