Conditional Cummulative Sum

Try this:

create this calculated column:

ColumnAUX = 
VAR _skuCncCost =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SKU|CNC|COSTURA] ) )
VAR _sequence =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] ) )
VAR _skuCnc =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SKU|CNC] ) )
VAR _pcb =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] ) )
VAR _tmp1 =
    CALCULATE (
        SUM ( ANALISIS[BAL. EXPORTAR] ),
        FILTER ( ALL ( ANALISIS ), ANALISIS[SKU|CNC|COSTURA] = _skuCncCost )
    )
VAR _tmp2 =
    CALCULATE (
        SUM ( 'DETALLE DE LOTES'[DOCENAS] ),
        FILTER (
            ALL ( 'DETALLE DE LOTES' ),
            'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] = _pcb
                && 'DETALLE DE LOTES'[SKU|CNC] = _skuCnc
                && 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] < _sequence
        )
    )
VAR _tmp3 =
    ADDCOLUMNS (
        var _skuCnc2=CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[SKU|CNC]))
        var _pcb2 = CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[PLANTA COSTURA BALANCE]))
        var _s2 = CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[SECUENCIA DE APLICACION]))
        return
        CALCULATETABLE('DETALLE DE LOTES',
        FILTER (
            all('DETALLE DE LOTES'),
            'DETALLE DE LOTES'[SKU|CNC] = _skuCnc2
                && 'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] = _pcb2
                && 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] < _s2
        )),
        "tesst",
            IF (
                _tmp1 + _tmp2
                    <= CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ) * -1,
                CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ),
                0
            )
    )
RETURN
    IF (
        _tmp1 + _tmp2
            <= CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ) * -1,
        CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ),
        0
    )

and this calculated column:

BALANCE PLANTA ACTUAL = 
VAR _skuCncCost =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SKU|CNC|COSTURA] ) )
VAR _sequence =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] ) )
VAR _skuCnc =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[SKU|CNC] ) )
VAR _pcb =
    CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] ) )
VAR _tmp1 =
    CALCULATE (
        SUM ( ANALISIS[BAL. EXPORTAR] ),
        FILTER ( ALL ( ANALISIS ), ANALISIS[SKU|CNC|COSTURA] = _skuCncCost )
    )
VAR _tmp2 =
    CALCULATE (
        SUM ( 'DETALLE DE LOTES'[DOCENAS] ),
        FILTER (
            ALL ( 'DETALLE DE LOTES' ),
            'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] = _pcb
                && 'DETALLE DE LOTES'[SKU|CNC] = _skuCnc
                && 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] < _sequence
        )
    )
VAR _tmp3 =
    ADDCOLUMNS (
        var _skuCnc2=CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[SKU|CNC]))
        var _pcb2 = CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[PLANTA COSTURA BALANCE]))
        var _s2 = CALCULATE(SELECTEDVALUE('DETALLE DE LOTES'[SECUENCIA DE APLICACION]))
        return
        CALCULATETABLE('DETALLE DE LOTES',
        FILTER (
            all('DETALLE DE LOTES'),
            'DETALLE DE LOTES'[SKU|CNC] = _skuCnc2
                && 'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] = _pcb2
                && 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] < _s2
        )),
        "tesst",
            IF (
                _tmp1 + _tmp2
                    <= CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ) * -1,
                CALCULATE ( SELECTEDVALUE ( 'DETALLE DE LOTES'[DOCENAS] ) ),
                0
            )
    )
RETURN
    _tmp1 + CALCULATE(SUM('DETALLE DE LOTES'[ColumnAUX]),filter(ALL('DETALLE DE LOTES'),  'DETALLE DE LOTES'[SKU|CNC] = _skuCnc
                && 'DETALLE DE LOTES'[PLANTA COSTURA BALANCE] = _pcb
                && 'DETALLE DE LOTES'[SECUENCIA DE APLICACION] < _sequence
        ))

I got it to work on the sample data you provided.